Do roulette wheels have bias?

Do roulette wheels have bias? The short answer

Yes, a physical roulette wheel can be biased, and some have been. Wear, a loose fret between pockets, a warped bowl or a slightly tilted table could, in principle, make some pockets catch the ball more often than others. The glossary entry on wheel bias gives the one-line definition.

In practice, three things make this almost irrelevant for a modern player. Casinos maintain and replace wheels. A bias big enough to matter has to be enormous compared with what a few thousand spins can reveal. And most online roulette is RNG software, where a wheel is only a picture. This page sets out the history, the numbers and the UK rules, and our casino questions hub has related answers.

What a biased roulette wheel actually is

On a fair single-zero wheel every pocket has the same 1 in 37 chance. A wheel is biased if, because of its physical condition, some pockets have a genuinely higher chance and others a lower one. That is different from a number simply appearing a lot in a short run, which happens on perfectly fair wheels all the time.

The distinction matters because the roulette wheel numbers are arranged to balance the wheel: red and black alternate, and equal-sized sections have roughly equal sums. A bias would show up as a physical section of the wheel, not a pattern of colours or odd and even.

A biased wheel is a fault in the equipment. A ‘hot number’ on a results board is just recent history. The two are often confused, and only the first could ever change the odds.

Biased roulette wheels in history

The Wikipedia roulette article (opens in a new tab) records several episodes. In 1873 Joseph Jagger exploited a biased wheel at Monte Carlo, the best-known case. In the 1930s some professionals sought out rigged wheels, which were not hard to find at the time. Richard Jarecki is said to have observed more than 10,000 spins of each wheel to look for flaws.

The British episode in 1982 is the turning point: the industry answered with low-profile wheels, which have shallower pocket separators. The history of roulette and our Monte Carlo history pages give more background on these stories.

How big would a roulette wheel bias need to be?

A straight-up bet pays 35 to 1. It breaks even only if the number wins 1 time in 36. A fair number wins 1 time in 37. So a pocket would need to hit about 2.78% of the time instead of 2.70% just to cancel the house edge, before making any profit.

That gap is tiny, and random variation is large. The standard deviation of hits on one number over N spins is the square root of N × (1/37) × (36/37). The table puts that into numbers:

At 1,000 spins the difference you are looking for is less than a sixth of the normal random swing. To make the gap twice the swing, a common rough test, you need about 187,000 spins of the same wheel. At the 60 to 80 games an hour Evolution quotes for its Auto Roulette, that is roughly 2,300 to 3,100 hours of results. Our page on variance and standard deviation explains the formula.

A real-data example: when a ‘hot’ number is probably luck

The Wizard of Odds roulette questions (opens in a new tab) include analysis of spin records readers sent in from physical wheels. In one, a number came up 278 times in 8,672 spins against 234.38 expected. That looks dramatic, about 2.9 standard deviations above average, yet the analysis put the chance of it being pure luck at 1 in 466.

There is a second trap. A wheel has 37 numbers, so if you look at all of them, one is very likely to look unusually high just by chance. Picking the busiest number after the event and calling it biased is the same mistake as believing in hot and cold numbers.

Even the 1913 run of 26 blacks at Monte Carlo was a fair-wheel event. It is the classic illustration of the gambler’s fallacy: players bet heavily on red, expecting it to be due, and lost.

Roulette wheel bias and UK rules

The Gambling Commission’s RTS 7 standard on random outcomes (opens in a new tab) requires results in remote games to be acceptably random and bans compensated (adaptive) games. Its guidance says RNG output should be uniformly distributed and the next result should be computationally infeasible to predict.

For mechanical devices such as a live wheel, the same guidance says the physical items used to choose a result should not be altered and players should not be able to touch them. For software that simulates a wheel, RTS 7C says the theoretical probabilities should match the real device. In short, a digital wheel must behave like a perfect 37-pocket wheel. Our pages on random number generators and how casino games are tested explain the checks.

Can online and live roulette wheels be biased?

RNG roulette, including Evolution’s First Person games, has no physical wheel. Results come from software, so there is no wear, tilt or loose fret to create a bias. A software fault would be a compliance failure, not a pattern a player could spot from a session.

Live roulette does use a real wheel. Evolution’s Lightning Roulette, for example, uses a physical single-zero wheel with the ball launched by an Auto Roulette mechanism, and its Auto Roulette tables use a fully automated precision wheel with no dealer. Studios have every reason to keep wheels in good order, and the RTS rules cover the equipment. The RNG against live roulette comparison sets out the practical differences.

What to do instead of hunting for a biased wheel

Looking for bias is a poor use of time and money. These steps do far more to control what roulette costs you:

1. Pick a single-zero table: 2.70% against 5.26% on a double-zero wheel.
2. If you bet on red, black, odd, even, high or low, look for La Partage, which cuts that edge to 1.35%.
3. Ignore hot and cold number boards; they show the past, not the future.
4. Set a deposit limit and a session budget before you start.
5. Treat any loss as the cost of the entertainment and never chase it.

Our page asking whether some roulette numbers are more common than others covers the same ground from the results-board angle, and setting deposit limits takes about a minute at any UK casino.

Where to play roulette at UK-licensed casinos

Playing at a UKGC-licensed casino means the games are covered by the RTS randomness rules above. Our partners’ welcome offers are not roulette offers: table and live games are often excluded from bonuses or weighted lower for wagering, so check the terms and look in each casino’s live and table lobby for single-zero games.

18+, full terms apply. Casumo offers a 50% bonus up to £100 on a first deposit of at least £20 (opt-in required), with 10x wagering on the bonus within 7 days, a £5 maximum bet while it is active, and game, play and payment restrictions. 10bet offers a 100% match up to £25 if you select the casino welcome bonus at sign-up and deposit within 7 days; it must be wagered 10x on slots within 3 days, bonus winnings are capped at £1,000, and Skrill and Neteller deposits are excluded.

Coral gives 100 free spins at 10p each on selected slots when you deposit and bet £10 or more on slots between 1 October 2026 and 1 February 2027. Accept the spins within 48 hours and use them within 7 days; winnings carry no wagering. Neteller, PayPal, Paysafe, Skrill, prepaid cards and certain debit cards don’t qualify.

18+. Every casino game has a built-in house edge, so over time you should expect to lose what you stake. Set a budget before you play and stop when it’s gone: safer gambling tools and support. Free, confidential help: National Gambling Helpline 0808 8020 133.