Home › Gambling in South Africa

Gambling in South Africa Statistics 2026: 90 Sourced Data Points Behind a R1.5 Trillion Headline

By the 16Best gambling data desk · Published 29 July 2026 · Last reviewed 29 July 2026 · About 30 minutes to read. Primary source: the National Gambling Board’s audited FY2024/25 statistics deck, published 9 October 2025, read alongside its Annual Report 2024/25.

Twenty rands staked for every one lost. That is what South Africa’s trillion-rand gambling headline actually describes. The National Gambling Board’s R1.50 trillion for FY2024/25 is turnover — money wagered, winnings included when they are wagered again — and operators kept R74.5 billion of it, a hold of 4.96%. That is the thinnest keep rate we have measured anywhere a regulator publishes both sides of the ledger, and it has thinned in each of the last three years. The trillion measures velocity, not value.

Every figure below is attributed to a named, dated source, and every number we worked out ourselves is stamped 16Best analysis and shown with its inputs so you can redo it. Sources are tiered where it matters: audited regulator returns, unaudited regulator prose, and non-representative panel surveys are not the same thing, and we say which is which.

How big is South Africa’s gambling market in 2026?

About R74.5 billion, not R1.5 trillion — and the gap between those two numbers is twenty-fold. The National Gambling Board publishes both in the same audited deck, defines both in its own glossary — and the press, the trade titles and the parliamentary write-ups almost all reach for the bigger one.

The NGB is explicit about what turnover means. It is “the rand value of money wagered,” and it “includes recycling, which refers to amounts that are staked on more than one occasion.” A punter who deposits R100, wins R95 back and stakes it again has generated R195 of turnover and R5 of revenue. Do that forty times over a Saturday afternoon and the turnover column reads thousands while the operator has kept the same R5. Gross gambling revenue is the other quantity: turnover less winnings paid out. It is what the industry earns, what the provinces tax, and what a market-size figure is supposed to mean.

Divide one by the other and South Africa’s implied hold for FY2024/25 is 4.96%. Put physically: of every R100 staked in South Africa last financial year, R95.04 went back out as winnings. Money crossed the national book at a rate of R47,586 every second of the 365-day year, and the industry kept R2,362 of it.

The scale error this produces is not subtle. At the 2024 calendar-year average of R18.33 to the dollar, R74.5 billion of GGR is about US$4.06 billion — roughly 0.6% of the world’s US$712 billion of gambling revenue, and less than a seventh of the US$30.9 billion Macau’s casinos won in calendar 2025 (MOP247.4 billion, converted at about MOP8 to the dollar). Read the turnover figure as revenue instead and South Africa becomes US$81.9 billion, or 11.5% of global gambling — about 2.6 times Macau. One country, two readings, an order of magnitude and change between them. Both Macau comparisons set a fiscal year ending 31 March 2025 against a calendar year, so treat them as scale, not as a league table.

Against South Africa’s own economy the same point lands harder. Nominal GDP in 2024 was about R7.3 trillion. The turnover figure equals 20.6% of everything the country produced; the revenue figure equals 1.02%. Any number that claims a fifth of national output has gone through one industry is measuring circulation, not consumption.

What did the numbers look like when casinos ran the market?

Fewer rands, staked far fewer times — South African gambling used to be a business with an ordinary margin. In FY2010/11 casinos took 81.9% of gross gambling revenue and betting 11.7%. The apparatus was physical, the product was slow, and the hold was whatever a slot floor holds.

Casinos still behave that way. Their implied hold has barely moved in three years: 5.80% in FY2022/23, 5.84% in FY2023/24, 5.68% in FY2024/25. Casino turnover has actually shrunk in nominal rands across that window, from R299.2 billion to R292.8 billion. And casino revenue peaked before the pandemic: the NGB puts pre-pandemic casino GGR at R18.4 billion in FY2019/20 and says the mode is still below it. It is. Casinos won R16.65 billion in FY2024/25 — 9.5% below their nominal peak six years earlier, before any inflation adjustment. Over the same six years national GGR went from R32.65 billion to R74.50 billion, so the whole industry is 2.3 times the size it was while its founding sector is smaller (16Best analysis of the NGB’s own per-province GGR series).

Limited payout machines are the useful control. Pub machines in South Africa hold 7.59% of the money that goes through them, the highest of any mode, and the figure has moved by fractions of a percentage point in three years. Bingo holds 6.28%. Neither has grown much. Both keep more of every rand than the mode that has swallowed the market.

When did betting overtake casinos, and what did it do to the margin?

FY2020/21, the hard-lockdown year — and the national hold has fallen in every year since. On the NGB’s fifteen-year share series casinos dropped from 56.3% of GGR in FY2019/20 to 39.2% in FY2020/21, while betting rose from 26.8% to 45.6% in the same single step. Betting never gave the lead back; by FY2024/25 it was 69.8% and casinos 22.3%. It compounded at 49.8% a year over the three years to FY2024/25, against 6.6% for casinos, 4.6% for bingo and 3.9% for LPMs.

Here is what nobody has put next to that. The mode that won is the mode that keeps the least, so the country’s aggregate margin has been falling the entire time the headline has been climbing.

Turnover and GGR as published by the National Gambling Board (FY2022/23 sector-performance summary; FY2023/24 and FY2024/25 audited statistics decks). Exact figures: turnover R559,880,253,827 / R815,110,216,812 / R1,142,683,922,400 / R1,500,660,968,064; GGR R34,430,989,724 / R47,168,024,499 / R59,306,767,610 / R74,498,495,925. Hold, the multiple, both CAGRs and the growth rates are 16Best calculations from those published totals. They match the NGB’s own printed growth figures in every step but one: its FY2022/23 executive summary rounds the GGR increase to 36.9% where the published totals give 37.0%. The FY2023/24 deck also narrates the previous year’s turnover as R814.0bn even though its own trend chart still prints R815,110 million; we use the figure each deck published for its own year, and the difference moves the FY2022/23 hold by less than one hundredth of a point.

Why is turnover growing faster than revenue?

Two reasons, and the smaller one is the obvious one. The obvious answer is mix: betting keeps less than casinos, betting got bigger, so the average fell. Decompose it properly and that explains barely a sixth of the move.

National hold fell 0.822 percentage points between FY2022/23 and FY2024/25, from 5.787% to 4.964%. Take FY2024/25’s turnover weights, apply FY2022/23’s mode-level holds to them, and you get 5.659% — so the shift toward betting accounts for 0.127pp, or 15.5% of the fall. The remaining 0.695pp, or 84.5%, came from margins falling inside the modes, overwhelmingly inside betting: bookmakers and totalisators went from holding 5.51% to 4.62% in two years. The decomposition is exact — there is no residual term — because the counterfactual holds one side fixed at a time (16Best analysis).

16Best analysis. Hold is GGR divided by turnover for each mode, computed from the NGB’s published mode tables for each year (FY2024/25 betting turnover R1,125,602,965,951 against GGR R51,974,933,051; casinos R292,840,156,194 against R16,645,815,707; LPMs R54,685,793,486 against R4,148,159,930; bingo R27,532,052,433 against R1,729,587,238). The FY2022/23 inputs behind the decomposition above are betting R430,994,922,371 against R23,748,237,393, casinos R299,211,100,905 against R17,342,258,570, LPMs R54,426,852,537 against R4,230,348,312 and bingo R30,477,340,999 against R1,847,180,224; the betting turnover is the residual of the published national total after the three modes the FY2022/23 summary prints separately, and the NGB’s own text puts it at “R431 billion… 52.9% of turnover.” Note the ordering: the largest mode by turnover has the smallest margin, and a pub machine keeps 64% more of every rand than a bookmaker does.

What sits behind the betting number is a product change the statistics do not label. The NGB records betting turnover in two buckets — horse racing, and “sport and other contingencies” — and the second bucket did R1,112.26 billion of the R1,125.60 billion. Horse racing, the original reason the bookmaker’s licence exists, took R13.34 billion — 1.2% of betting turnover. The category that grew is the one whose name refuses to say what is in it.

Online is the visible half of that. The NGB puts online betting GGR at R44.46 billion in FY2024/25 against R28.97 billion in FY2023/24 — up 53.5% in a year, though the comparison is slightly generous because the disaggregating provinces went from five to six. The province that joined was the Eastern Cape, and its entire betting revenue for the year was R188 million against a national increase of R15.5 billion, so the wider net accounts for at most a rounding of that growth (16Best analysis). Online betting alone is now 59.7% of all South African gambling revenue, which is the point our sibling page on South African gambling law builds its case on.

Which provinces actually keep the money?

Almost none of them — two provinces run the handle, and they run it at the thinnest margins in the country. Mpumalanga booked R623.0 billion of turnover (41.5% of the national total) and the Western Cape R402.5 billion (26.8%), which is 68.3% between them, from provinces holding 20.1% of South Africa’s people. On betting alone the concentration is starker: R608.7 billion and R343.5 billion of a R1.126 trillion book, or 84.6%. Gauteng, the province with a quarter of the population and the largest casino floors in the country, books 12.0% of national turnover.

Now compute what each province’s bookmakers actually keep, and the country stops looking like one market.

Turnover and GGR summed by 16Best from the NGB’s quarterly per-province betting tables for FY2024/25; the nine provincial turnover figures reconcile exactly to the published national total of R1,125,602,965,951, and the nine GGR figures to R51,974,933,051. Both hold columns are ours. Column four divides GGR by turnover. Column five takes the NGB’s own quarterly return-to-player percentage for betting in each province, weights the four quarters by that province’s quarterly turnover, and subtracts from 100. The NGB’s national betting RTP of 92% is an unweighted average across the nine provinces, which is why it does not equal 100 minus the national hold; weighted by turnover it is 95.45%.

This is the mechanism behind the whole page. South Africa’s national hold is low because the national handle is overwhelmingly booked in the two provinces whose licensees run the highest-return, fastest-recycling product in the country. As those provinces take more share, the country’s implied margin falls further, and the trillion-rand headline accelerates away from the revenue underneath it.

What does the average South African adult stake and lose?

About R34,669 staked and R1,721 lost across the year — R95 a day going in, R4.72 a day staying in. Stats SA put the mid-2025 population at 63,100,945, of whom about 43.3 million are 18 or older (we exclude the 0–14 cohort in full and three-fifths of the 15–19 band; the pro-rating is ours).

16Best analysis. Adult population derived from Statistics South Africa Mid-Year Population Estimates 2025 (P0302). Dollar figures use the 2024 calendar-year average of R18.3346 per US$, which covers nine of the twelve months of FY2024/25; at the 2025 average of R17.88 the per-adult loss reads US$96, and at the R17.35 implied by CDC Gaming’s October 2025 conversion it reads US$99. GDP is Stats SA nominal GDP at market prices for calendar 2024, set against a fiscal year running April 2024 to March 2025, so read it as an order of magnitude.

Regionally the contrast runs the other way. Kenya’s estimated bookmaking GGR of about US$354 million across roughly 32.8 million adults is near US$11 per adult; and Nigeria’s gambling revenue works out near US$726 million on the central case in our Nigeria analysis — a derived figure at an assumed 20% hold on an online-market handle that has lottery inside it, not an audited one, because Nigeria publishes no handle. Even taken at face value it is under a fifth of South Africa’s, from a population several times larger. South Africa is where the money is on the continent. It is simply nowhere near where the turnover headline suggests.

What does the state collect per rand staked?

39 cents per R100 staked — and 7.80% of what operators actually keep. The nine provinces collected R5,812,135,921 in FY2024/25: R3.42 billion from betting, R1.72 billion from casinos, R525 million from LPMs and R146 million from bingo.

Measured against GGR, those collections imply effective rates of 10.3% on casinos, 12.7% on LPMs, 8.4% on bingo and 6.6% on betting — the schedule our South African gambling laws page sets out in full. Measured against turnover, which is the yardstick this page is built on, the picture is different and worth stating on its own.

16Best analysis of NGB tax, GGR and turnover tables, FY2024/25. The GGR column is an implied national average across nine different provincial schedules, not a statutory rate, and excludes corporate income tax and VAT. The turnover column is a derived ratio for comparison only — no South African province taxes turnover.

The spread is the story. A pub machine hands the state 3.2 times as much per rand wagered as a bookmaker does, on a base one twentieth the size. National Treasury’s discussion paper of 25 November 2025 proposes a 20% national tax on online and interactive gross gambling revenue — levied on top of existing provincial rates, which the paper says would take the all-in charge on online gambling to between 26% and 29%. Applied to the FY2024/25 online betting GGR of R44.46 billion, 20% is about R8.9 billion, more than half again what all nine provinces collect from all four modes today (16Best analysis). Treasury says the measure is expected to raise more than R10 billion, which implies a taxable base above R50 billion — roughly R5.5 billion more online revenue than the NGB can currently disaggregate, and a different R50 billion entirely from the illegal-market estimate we refuse to use below. It is a proposal, not law: comments closed on 27 February 2026 after an extension from 30 January, and nothing has been enacted.

What do the participation and problem-gambling numbers really say?

Less than the headlines built on them, and the two most-quoted figures need their denominators stated. This section is deliberately narrow: we report what the instruments measured and who they measured it on.

The NGB’s Annual Report 2024/25 states that its socio-economic research found “increases in gambling prevalence nationwide coupled with an increase in problem gambling to 31%.” That is a single sentence in the report’s overview. There is no instrument named, no sample frame published, no denominator stated, and no underlying study released. What is published sits one step removed: press coverage of the same research, in October 2025, reports a sample of about 4,000 respondents and a gambling-participation rate of 65.7%, and describes the 31% as a share of those who gambled rather than of everyone asked. On that reading the two numbers multiply to roughly 20% of the whole sample — a materially different claim from “31% of South Africans”, which is how the figure usually travels (16Best analysis of the two press-reported figures; the NGB itself publishes neither the multiplication nor the denominators). Until the study is released, the honest version is: 31% of gamblers in one unreleased 4,000-person survey. It cannot be set against South Africa’s older National Responsible Gambling Programme prevalence work, which used different instruments in a different decade and produced figures in the low single digits.

The second figure travels even further from its base. GeoPoll’s April 2025 fieldwork across six African countries found 83% of its South African respondents had placed a bet — the highest of the six. That sample was 68% male and 80% aged 18–34, and GeoPoll describes it as not fully demographically representative. It is a real measurement of a young, male, online panel. It is not a national participation rate, and multiplying it by the population produces a bettor headcount that means nothing.

What is not in dispute is demand on the support infrastructure. The counselling line run by the South African Responsible Gambling Foundation logged 1,013,626 calls in FY2024/25 against 140,263 the year before, a 623% increase. Read that carefully too: 48,710 of the calls were wrong numbers — people phoning an addiction helpline to cash out, close an account or upload FICA documents — and that number actually fell from 67,919, because the Foundation installed a diversion menu to push them back to operators. Referrals rose 55% to 4,166: 3,090 problem gamblers, 902 self-exclusions, 134 relapses and 40 family referrals.

The funding tells its own story. The industry levy for the Foundation is capped at 0.1% of GGR, and the NGB prints its own actual-versus-expected table: against the FY2023/24 national GGR of R59.3 billion, expected collections were R59,306,768. The Foundation received R40,947,861 — a record, and still 31% short of the formula (the shortfall is the NGB’s own figure; the percentage is 16Best analysis. The Foundation’s numbers are printed unaudited in the annual report, with audited versions promised later). A country whose gambling turnover grew 31% in a year funds its treatment programme on roughly one twentieth of one percent of what its operators keep. South Africa’s minimum gambling age is 18, and free counselling is available through the National Responsible Gambling Programme.

How does South Africa’s hold compare with other markets?

It is the lowest we have measured from published regulator returns: South African bookmakers keep 4.62% of turnover against 10.15% for American sportsbooks, so a rand takes 21.7 stakes to become revenue where a dollar takes 9.9. That is the closest like-for-like pair available — regulated sports betting against regulated sports betting — and it is still not identical, because South Africa’s year ends 31 March 2025 and the American figure is calendar 2025. Every scope in the chart below is stated on the chart, and the one bar that is not a measurement is labelled as an assumption.

A rand staked with a South African bookmaker earns its operator 4.6 cents. A dollar staked with an American sportsbook earned 10.1 cents in calendar 2025, on US$165.58 billion of handle against US$16.80 billion of revenue. Same broad product, less than half the margin. Kenya after its 2019 crackdown ran at 27.4% — a figure that carries public lotteries inside it, so it is not a clean bookmaker number — because the high-volume, thin-margin product left with the operators that ran it; South Africa is the mirror image of that market, and the full country-by-country picture sits in our sports betting laws by country hub.

Why do South African gambling figures disagree?

Because the country publishes four different big numbers, one regulator prints two versions of the same year, one large channel is measured in only six of nine provinces, and one activity cannot be measured at all. In the order they produce wrong sentences:

• Turnover is not revenue, and the multiple is not stable. This is the error everyone makes, but the version worth knowing is the second-order one: the conversion factor between the two numbers is itself moving. R1.50 trillion staked and R74.5 billion kept is 20.1 to 1 in FY2024/25; it was 16.3 to 1 three years earlier. So a reader who learned the rule of thumb “divide South African turnover by about sixteen” in 2022 and applied it in 2025 would overstate the market by 23%. There is no fixed divisor. Take the GGR the NGB publishes or take nothing, and never set a turnover figure against another country’s revenue — Macau’s US$30.9 billion in calendar 2025, global US$712 billion — because that is a river measured against a reservoir.
• Hold and return-to-player are not complements of the same thing, even though here they nearly are — if you weight them. The NGB’s published RTP is measured across game plays and printed quarterly by province; our hold is annual revenue over annual turnover across recycled stakes. Weight the four quarterly RTPs by each province’s quarterly turnover and the two measures land within 0.22 of a point everywhere. Take the NGB’s national betting RTP of 92% at face value and you get a hold of 8%, which is wrong by a factor of nearly two — because that 92% is an unweighted average of nine provinces, six of which are rounding errors in the national book. Average the nine provincial figures flat and you get 91.6%, which is what the deck prints as 92%. Weight them by turnover and you get 95.45%, and 100 minus that is 4.55% against our 4.62%. This is the single easiest mistake to make with this dataset.
• The NGB publishes two totals for FY2024/25. The audited statistics deck gives R74,498,495,925; the Annual Report 2024/25 gives R74,857,617,238 under a footnote reading “audited statistics received from all PLAs excluding Mpumalanga.” The entire R359,121,312 difference sits in betting — R51,974,933,051 against R52,334,054,363 — which is an unaudited Mpumalanga, the province that books 41.5% of national turnover. We use the statistics deck throughout, as National Treasury does (16Best analysis: the difference is ours to locate, the two totals are the NGB’s).
• The audited deck is not internally consistent about its own shares. Its mode pie prints betting at 69.8% and casinos at 22.3%; its fifteen-year trend chart prints 69.9% and 22.2% for the identical year. Neither the deck total nor the annual-report total reproduces that second pair — 22.2% needs a base of about R75.0bn and 69.9% a base of about R74.4bn, and no such pair exists. It is a rounding-and-vintage artefact, not two defensible bases. Compute mode shares from the rand figures and ignore the printed percentages (16Best analysis).
• Nine provincial regulators, one national compiler, and a past that keeps moving. Every figure here originates with a provincial licensing authority and is aggregated by the NGB, and provinces differ in what they require licensees to report and how quickly. So prior years get restated. FY2022/23 turnover appears as R815.1bn in its own summary; the following year’s deck narrates it as R814.0bn while that same deck’s trend chart still prints R815,110 million. Gauteng is the sharpest case: the FY2024/25 deck restates Gauteng’s FY2022/23 GGR down by R1.008bn (R12,776,979,379 to R11,769,098,275) and its FY2023/24 GGR up by R520m (R13,095,261,666 to R13,615,681,141), with no other province touched. The consequence is arithmetic: in the FY2024/25 deck’s per-province chart the FY2024/25 column sums exactly to R74,498,495,925, but the FY2022/23 column sums to R46.16bn against the R47.17bn the NGB published for that year, and the FY2023/24 column to R59.83bn against R59.31bn. Take each year from its own deck (16Best analysis; both decks are linked below).
• “Online” is a partial measurement. Only the Western Cape, Mpumalanga, Limpopo, North West, Eastern Cape and Northern Cape disaggregate online from retail betting. Bets placed online in Gauteng, KwaZulu-Natal and the Free State fall into the retail column. So R44.46 billion and 59.7% of GGR are floors. The set also changed between years — five provinces in FY2023/24, six in FY2024/25 — which flatters year-on-year online growth by a small amount.
• Online casino volume is unmeasurable by construction. Interactive gambling has no South African licence, therefore no returns and no reporting obligation. A figure of about R50 billion a year for the illegal offshore market circulates in 2026 coverage; we cannot trace it to a published methodology and do not adopt it. Any number for a prohibited market is an estimate resting on assumptions.
• The National Lottery is outside every figure on this page. Lotteries and sports pools are licensed nationally under the Lotteries Act 57 of 1997 and appear nowhere in NGB statistics. Lottery ticket sales are also turnover, not revenue — the “R180 billion licence” figure quoted during the 2026 operator handover is projected eight-year ticket sales.
• Rands and dollars across a moving rate. The rand averaged R18.33 to the dollar in 2024 and R17.88 in 2025; CDC Gaming’s widely cited US$86.5 billion turnover conversion implies about R17.35, an October 2025 rate applied to a year that ended in March. The same R74.5 billion reads US$4.06bn, US$4.17bn or US$4.29bn depending on which you pick. Cite the rand; state the rate and date for any dollar.

Who produces these numbers, and which way do they lean?

Licensees do. Every rand on this page began life as an operator’s own return to a provincial regulator, and nobody counts a South African stake independently. That is not a scandal — it is how gambling statistics work almost everywhere — but it decides which figures to lean on and which to hold loosely.

Trace the chain. A bookmaker files turnover, winnings paid and gross revenue with its provincial licensing authority. The authority audits, assesses tax, and passes totals to the National Gambling Board, which compiles rather than re-measures. There is no meter room, no clearing house, no third-party settlement feed. Which means the reliability of a number depends less on the NGB than on which provincial audit it survived — and in the annual-report version of FY2024/25, the province carrying 41.5% of national turnover had not been audited at all.

Now ask what each number is for, because that is where the lean lives. Turnover is nobody’s tax base in South Africa; no province levies on it. It is the number that makes the sector look consequential in a parliamentary presentation, and it is the number the trade press reprints. Gross gambling revenue is the taxed quantity in all nine provinces. The two numbers therefore sit under different pressures and different audit intensities, and this page is built entirely on their ratio. We have no evidence of misreporting in either direction and make no such allegation. We do note that the one quantity everybody quotes is the one nobody is taxed on (16Best analysis).

One more asymmetry worth naming: the online share is structurally understated and will keep being understated for reasons that have nothing to do with gambling behaviour. Three provinces — Gauteng, KwaZulu-Natal and the Free State — do not require licensees to split online from retail, so every online bet placed under those licences is counted as retail. When a fourth province starts disaggregating, the “online” line will jump and every outlet will report growth. Some of it will be growth. Some of it will be a reporting rule changing.

Back to top

Key takeaways

• The trillion measures velocity, not value. R1.50 trillion staked, R74.5 billion kept, a 4.96% hold, and every rand lost staked 20.1 times before it was gone. Anyone comparing that trillion to another country’s gambling revenue is out by more than an order of magnitude.
• The gap widens every year. 16.3× in FY2021/22, then 17.3×, 19.3×, 20.1×. Turnover compounded at 38.9% a year while revenue compounded at 29.3%. The headline is accelerating away from the economics beneath it.
• Mix is the small half of the explanation. Only 15.5% of the two-year margin compression came from betting taking share. The other 84.5% came from margins falling inside the modes, chiefly betting’s own drop from 5.51% to 4.62%.
• The country does not have one hold; it has nine. Gauteng bookmakers keep 15.4% of stakes and Mpumalanga bookmakers 3.5% — and the thin end is where 84.6% of the national betting handle is booked. The NGB’s own return-to-player figures confirm the spread independently.
• Per adult, South Africa is not an outlier. About R1,721 lost per adult a year, near US$94, below the roughly US$132 global average — before the National Lottery, which sits outside these figures. A high-velocity market is not the same thing as a high-loss market.
• Tax lands on the shrinking quantity. Provinces take 7.80% of GGR but 39 cents per R100 staked, and just 30 cents on betting against 96 cents on a pub machine. A 20% national online tax is proposed on the one measure that keeps falling as a share of the headline.
• Read the harm data at its own denominator. The NGB’s 31% problem-gambling figure comes from one sentence of prose about an unreleased survey with an unnamed instrument; on the press reading it describes gamblers, not adults, and lands near 20% of the sample. GeoPoll’s 83% describes a young, male, online panel. The 1,013,626 helpline calls are a counted fact and the more useful indicator.
• Nobody measures a South African stake independently. Every figure here is an operator return, audited by a provincial authority and compiled by the NGB. The quantity everyone quotes — turnover — is the one no province taxes. That is worth remembering before treating it as a market size.

Back to top

Frequently asked questions

How much do South Africans gamble each year?

South Africans staked R1.50 trillion in the financial year ending 31 March 2025, according to the National Gambling Board's audited statistics published on 9 October 2025. That figure is turnover, meaning the total amount wagered including winnings that were staked again, and it is not the size of the market. Gross gambling revenue, which is what operators actually kept after paying winnings, was R74,498,495,925. Dividing revenue by turnover gives an implied national hold of 4.96 percent, which means about R95.04 of every R100 staked was returned to players as winnings.

Is the South African gambling market really worth R1.5 trillion?

No. R1.5 trillion is turnover, not revenue. The operator revenue figure for the same year is R74.5 billion, roughly one twentieth of it. The National Gambling Board's own glossary defines turnover as the rand value of money wagered including recycling, meaning amounts staked on more than one occasion. Comparing the R1.5 trillion turnover figure with another country's gross gambling revenue, or with a global market-size estimate, overstates South Africa by more than a factor of twenty.

What is the gambling hold rate in South Africa?

The implied national hold for the 2024/25 financial year was 4.96 percent across all four regulated modes. By mode it was 7.59 percent for limited payout machines, 6.28 percent for bingo, 5.68 percent for casinos and 4.62 percent for bookmakers and totalisators. The national figure has fallen in every year since 2021/22, when it was 6.15 percent. It varies enormously by province within betting, from 15.39 percent in Gauteng to 3.50 percent in Mpumalanga, and the National Gambling Board's separately published return-to-player figures corroborate that spread. Hold is not the same thing as house edge: turnover includes winnings that were staked again, so this is an annual keep rate across recycled stakes.

Which province has the most gambling in South Africa?

Mpumalanga on turnover, the Western Cape on revenue. Mpumalanga booked R623.0 billion of turnover in 2024/25, or 41.5 percent of the national total, against the Western Cape's R402.5 billion and Gauteng's R179.7 billion. On gross gambling revenue the Western Cape led with R23.13 billion, or 31.0 percent, followed by Mpumalanga at R22.25 billion and Gauteng at R13.49 billion. The reason for the mismatch is margin: Mpumalanga's bookmakers keep about 3.5 percent of every rand staked while Gauteng's keep about 15.4 percent. Between them Mpumalanga and the Western Cape account for 68.3 percent of national gambling turnover, but 60.9 percent of national gambling revenue, while holding about 20 percent of the population.

How much tax does South African gambling pay?

The nine provincial licensing authorities collected R5,812,135,921 in gambling taxes and levies in the 2024/25 financial year. That is 7.80 percent of gross gambling revenue, or about 39 cents for every R100 staked. By mode the implied effective rates on revenue were 12.66 percent for limited payout machines, 10.32 percent for casinos, 8.42 percent for bingo and 6.59 percent for betting. There is no national gambling tax. National Treasury proposed a 20 percent national tax on online and interactive gross gambling revenue in a discussion paper published in November 2025, with comments closing on 27 February 2026, but it has not been enacted.

How much does the average South African lose to gambling?

About R1,721 per adult across the 2024/25 financial year, or roughly R4.72 a day, derived by dividing R74.5 billion of gross gambling revenue by an estimated 43.3 million South Africans aged 18 and over based on Statistics South Africa's mid-2025 population estimates. At the 2024 average exchange rate of R18.33 to the US dollar that is about US$94, which is below the global average of roughly US$132 lost per adult per year. Two caveats matter: the figure excludes the National Lottery, which sits outside National Gambling Board reporting, and an average across all adults says nothing about how losses are distributed among people who actually gamble.

How many South Africans gamble?

There is no reliable population-level count. The National Gambling Board's annual report for 2024/25 refers to rising gambling prevalence and to problem gambling increasing to 31 percent, without naming the screening instrument, the sample frame or the denominator, and the underlying socio-economic study has not been published. Press coverage of the same research reports a sample of about 4,000 respondents and a gambling-participation rate of 65.7 percent, and describes the 31 percent as a share of those who gambled rather than of everyone surveyed, which would put it near 20 percent of the sample. Separately, GeoPoll's April 2025 fieldwork found 83 percent of its South African respondents had placed a bet, the highest of six African countries surveyed, but that sample was 68 percent male and 80 percent aged 18 to 34 and GeoPoll states it was not fully demographically representative. Neither figure should be multiplied by the population to produce a bettor headcount.

Sources

• National Gambling Board — National Gambling Statistics FY2024/25 (audited, published 9 October 2025) (turnover R1,500,660,968,064 and GGR R74,498,495,925 by mode and province; quarterly per-province betting turnover and GGR; taxes and levies R5,812,135,921; published return-to-player by mode and province; 15-year GGR share series). Audited regulator return — the highest-confidence tier on this page, and the source of every rand figure unless stated otherwise.
• National Gambling Board — National Gambling Statistics FY2023/24 (audited) (turnover R1,142,683,922,400 by mode; GGR R59,306,767,610; online betting GGR R28,969,970,069 across five disaggregating provinces)
• National Gambling Board — Gambling Sector Performance in South Africa, FY2022/23 summary (turnover R815,110,216,812 by mode; GGR R47.2bn; FY2021/22 turnover of R559.9bn; pre-pandemic casino GGR peak of R18.4bn; glossary definition of turnover and recycling)
• National Gambling Board — Annual Report 2024/25 (socio-economic research and the 31% problem-gambling statement; helpline volumes of 1,013,626 calls against 140,263; referral breakdown; employment 33,169; alternative GGR total of R74,857,617,238 excluding an audited Mpumalanga). Regulator narrative report; the Foundation’s figures inside it are printed unaudited by the NGB’s own footnote.
• Statistics South Africa — Mid-Year Population Estimates 2025 (P0302) (total population 63,100,945; single age-band table used to derive the 18-plus adult base; Western Cape 7,627,688 and Mpumalanga 5,076,133)
• Statistics South Africa — Gross domestic product, fourth quarter 2024 (nominal GDP at market prices of about R7.3 trillion for calendar 2024)
• National Treasury — The Case for a National Online Gambling Tax (discussion paper, November 2025) (proposed 20% on online and interactive GGR; provincial rate ranges; comment deadline 27 February 2026)
• ICLG — Gambling Laws and Regulations South Africa 2026 (the Western Cape and Mpumalanga regulators as the readiest to permit casino-style contingencies under bookmaker licences; provincial tax schedules)
• GeoPoll — Report: Betting in Africa 2025 (April 2025 fieldwork, 4,191 respondents across six countries; South Africa 83%, the highest of the six; sample 68% male and 80% aged 18–34, described as not fully demographically representative). Non-probability online panel — a real measurement of a specific population, not of South Africa.
• CDC Gaming — Online betting drives South Africa’s US$86.5B gambling turnover last year (the October 2025 conversion implying about R17.35 to the dollar)
• iGaming Business — South Africa gambling revenue up to ZAR75bn in 2025, parliamentary committee flags black market threat (NGB presentation to the Portfolio Committee on Trade, Industry and Competition, 17 October 2025; direct employment of 33,169; provincial tax split)
• Casino News Daily — Online gambling boom raises alarms in South Africa, 20 October 2025, and TimesLIVE — Gambling obsession is costing SA dearly, 18 October 2025 (press reporting of the NGB research: gambling participation of 65.7% and the 31% read as a share of those gamblers; sample of about 4,000; helpline volumes). Press summary of an unpublished study — the weakest tier on this page, and flagged as such wherever it is used.
• Exchange-Rates.org and Exchange Rates UK — USD to ZAR history (2024 annual average of about R18.33 per US$, taken as R18.3346; 2025 annual average of about R17.88, taken as R17.8782 — the rates used for every dollar conversion on this page)
• RG.org — US sports-betting tracker aggregating state regulator filings, as set out in how much Americans bet on sports (US$165.58bn of legal sports-betting handle against US$16.80bn of revenue, a 10.15% hold, calendar 2025). Real-time tracker, not an audited annual return.
• iGaming Business — Sizekhaya takes South Africa’s R180 billion lottery licence as court battle looms (the R180bn attached to the 2026 National Lottery handover is projected ticket sales across an eight-year licence, not revenue and not a National Gambling Board figure)
• Yogonet and Casino.org reporting of Macau Gaming Inspection and Coordination Bureau data — Macau casino revenue hits US$30.9 billion in 2025 (MOP247.40bn for calendar 2025, up 9.1%)
• H2 Gambling Capital — global gross gambling revenue of about US$712bn and player losses of about US$573bn in 2024, the benchmarks used in how much money casinos make and our gambling loss statistics. Vendor estimate, not a regulator return. The Nigerian comparison of about US$726m is not H2GC and not a measurement: it is our own central case at an assumed 20% hold, set out in Nigeria gambling statistics, because Nigeria publishes no audited handle.

Back to top