The slot RTP database: how often slots really pay

the lab · updated 2026-07-03 · simulation-based

What a 96% RTP actually does to $100

A 96% RTP does not mean you get $96 back from $100. It means that over an enormous number of spins the game returns 96% of everything wagered, and a session is not an enormous number of spins. Our standard measurement is $0.50 a spin against a $100 bankroll, 10,000 sessions per configuration, capped at 2,000 spins. Under those conditions, on high-volatility slots with published RTPs between 96% and 96.6%, between 71% and 85% of sessions busted, the bankroll hit zero before the cap. The median busted session on these games typically ended with $0.31 to $0.36 of the original $100.

The same wager on gentler math looks different: Blood Suckers (98.00% published RTP, low volatility) busted only 27.4% of $0.50/$100 sessions, and Huff N' More Puff (96.00%) busted 44.4% with a median ending of $32.53. RTP matters, but volatility decides how the RTP reaches you, which is why we measure sessions instead of quoting spec sheets. Definitions with data: what RTP is, what a bust rate is, what volatility does.

The database: measured session outcomes, slot by slot

Eighteen flagship studies below; the full library is on the studies page. "Bust rate" is the share of $0.50-a-spin, $100-bankroll sessions that hit zero before the 2,000-spin cap, at the provider-default RTP. "Lowest version" is the lowest RTP configuration documented for that title in our files, the ladder a casino can legally choose from. "Top version seen at" links our review of the operator a dated scan sighted running the highest published version, only where we have verified it.

Read it the honest way: a 99% game and a 96% game are both negative-expectation, the difference is how long the money lasts and how the returns are distributed, not whether the house keeps a share.

How often do slots pay? Hit frequency is not payback

Slots pay something constantly and pay back rarely, the two numbers players conflate most:

• Sugar Rush lands a win roughly one spin in three (34.48% hit frequency), and still busted 82.0% of $0.50/$100 sessions, because the average win is far smaller than the average bet run.
• Book of Dead hits about 1 in 3.8 spins (26.2%); its free spins arrived, in our model, and paid a median 44.4x against a mean of 81.2x, the mean is dragged up by rare big bonuses most sessions never see.
• Dead or Alive 2's bonus arrived about once in 195 spins, and 68% of bonuses paid under 50x. Its famous 111,111x maximum win? Our simulation landed it once in 142,194,762 spins.
• In Wanted Dead or a Wild, the bonus everyone queues the game up for arrived once every 238 spins on average, at $0.50 a spin, that's about $119 wagered per bonus, against a $100 bankroll.
• In our Gates of Olympus experiment, doubling the bonus frequency at fixed RTP improved zero of nine measured cells, more frequent bonuses were paid for by a thinner base game. Frequency is a feel, not a payout rate.

More on the measurement: bonus frequency, defined with data.

The version lottery: the same slot pays less at most casinos

Most modern slots ship in several RTP versions, and the casino, not the game, picks which one you get. The published ladders in our files span up to 12 points: Moon Princess runs from 96.50% down to 84.50% across five steps, among the widest spreads in our library; Book of Dead's published ladder runs 96.21% to 84.18%.

Deployment scans show the lower rungs are not theoretical:

• Book of Dead: in a July 2026 scan of 20 casinos, none was confirmed running the 96.21% default; 16 were confirmed on lower versions (12 of them the 94.25% version).
• Reactoonz: of 24 scanned casinos, only one ran the 96.51% version every review quotes; observed versions ran down to an 87.50% floor.
• Money Train 2: 29 of 43 scanned casinos ran the lower 94.00% version instead of the 96.40% default.
• Legacy of Dead: 12 of 16 scanned casinos displayed the 94.51% version rather than the 96.58% default.

What a lower version costs, measured

We simulate the lower rungs instead of guessing. At $0.50/$100, moving from default to the documented floor raised the bust rate by +5.4 points on Wanted Dead or a Wild (88.42% version), +10.4 on Book of Dead (87.25%), +11.0 on Fire in the Hole 3 (84.07%), +14.6 to +24.6 across the grid on Moon Princess (84.50%), and +19 to +27 on Reactoonz (87.50%). Even the modal deployed version, the one most casinos actually run, costs real sessions: Reactoonz at 94.51% (the rung 14 of 24 scanned casinos ran) added about 5 points of bust rate and cut the median session by 175 spins versus the quoted default.

Nothing on screen tells you which version you're playing unless you open the game's info panel. How RTP versions work explains where to look; our casino pages track which operators we've observed running the highest published versions of studied slots.

Bonus buys: what a purchased feature actually returns

We also simulate feature buys, 100,000 buys per slot across 50 bonus-buy studies. The pattern is consistent: the average buy returns roughly the game's RTP, but the typical (median) buy returns a fraction of its cost, and most buys lose money, the average is carried by rare big hits. At the harsh end, San Quentin xWays returned less than the buy cost on 87.0% of buys, with a median return of 0.09x. The most extreme structure we've measured is Mental 2's 6,666x "God Mode", one spin whose published outcome set is 99,999x or nothing: it paid zero on 93.6% of simulated buys. On Fire in the Hole 3, where the provider publishes no buy RTPs at all, 72 to 74% of buys lost money under every assumption we tested.

The most and least survivable slots we have measured

Same wager, same bankroll, wildly different rides. The gentlest game in the library, Blood Suckers at 98% RTP (27.4% bust), and Mega Joker's 99% max-bet Supermeter mode (39.7%) sit roughly sixty points of bust rate away from the harshest measured cell, Zeus vs Hades at 84.6%. In between, the pattern is stable: classic low-volatility engines (Gonzo's Quest, 55.1%) survive longer; modern extreme-volatility engines concentrate the return into rare events and bust three-quarters or more of sessions. And the honest footnote on every row: the survivable ones lose money too, just more slowly. Mega Joker's 99% applies only at max-bet Supermeter play, and is still negative-expectation.

How we measure (and what we refuse to claim)

Every study starts from the provider's published math, RTP, hit frequency, volatility class, bonus structure. Those published numbers are inputs, not findings: we don't verify a provider's certification, we test what their published configuration does to sessions. Models are validated analytically and against 10M-spin runs before a study publishes (methodology). Scans of deployed versions are attributed to their source and date; where a figure is single-source or modelled, the study says so.

What this database cannot do, and will never claim to do: it cannot help you win. Every configuration we have ever simulated, including the 98 to 99% RTP games, is negative-expectation. The data tells you what a game is likely to cost and how it behaves; the only winning number on this page is the price of the entertainment. If play stops being entertainment, see our responsible gambling page. 18+.

FAQ

Explore the database

• All 101 lab studies, every slot, hundreds of millions of spins each
• Bonus-buy studies, 100,000 simulated buys per slot
• RTP versions, explained · RTP · bust rate · volatility · bonus frequency
• Which casinos run the highest published versions
• Methodology, how models are calibrated and validated

Simulation-based observations, not predictions. We never advise betting. 18+. Worldwide excluding the US.