On 18 August 1913, at the Monte Carlo Casino, the ball landed on black 26 times in a row. Players who were sure red was due lost heavily trying to catch the turn. Nobody had tampered with the wheel. A run that long has a chance of about 1 in 137 million from any given spin on a single-zero wheel, and with enough spins happening across enough tables, rare things do happen.
That story is the right place to start the question “is roulette rigged?”, because most suspicion comes from streaks, and streaks prove nothing on their own. What does matter is who checks the game, how, and what happens when a physical wheel develops a fault. This article covers each layer. For the odds a fair game should deliver, see roulette odds and payouts.
What a fair roulette game has to do
A fair game has two jobs. Each pocket must be equally likely on every spin, and no spin may depend on the ones before it or on what players have bet. If both hold, the results converge on the expected frequencies: on a European wheel each number comes up about once in 37 spins over a long run, each colour about 18 times in 37.
The payouts are a separate matter. They are set to give the house a 2.70 % edge on a single-zero wheel, and that edge is printed in the maths, not hidden. A fair game still costs money to play. “Rigged” would mean the outcome is not what the published odds say, and that is what testing is there to rule out.
How online roulette is tested
An online roulette game that is not streamed from a studio runs on a random number generator (RNG). The RNG produces a stream of values, and the game software maps each value to a pocket. Two things can go wrong: the stream could be predictable, or the mapping could favour some pockets.
Independent testing labs check both. In broad terms they review the source code, run the RNG through statistical test batteries over very large samples, confirm that the mapping gives each pocket its correct share, and check that the paytable in the game matches the maths the operator declared. A version that passes is certified; a changed version has to be tested again.
Three names come up most often:
- Gaming Laboratories International (GLI), founded in New Jersey in 1989, tests gaming equipment and software for regulators worldwide and publishes technical standards, including GLI-19 for interactive gaming systems.
- eCOGRA, a London-based testing and inspection agency known for its “eCOGRA Safe and Fair” seal.
- iTech Labs, an Australian-based lab that tests RNGs and online games for a range of jurisdictions.
A lab seal in a footer is a signal, not proof. The useful check is the licence, because the regulator decides which labs and which standards count.
What regulators in Canada require
In Ontario, the Alcohol and Gaming Commission of Ontario (AGCO) sets the Registrar’s Standards for Internet Gaming. Operators registered for the Ontario market have to offer games that meet those standards, which cover game fairness and the testing behind it. iGaming Ontario publishes the list of operators in its regulated market, and gaming suppliers also register with the AGCO. If you play in Ontario, our Ontario roulette guide lists only operators on that list.
Outside those two provinces, operators carry licences from authorities named in their listings, and the standards depend on that authority. Our scores only include casinos whose licence we have confirmed for the province you are in.
Live dealer wheels in studios
A live roulette table is a physical wheel, a real ball and a dealer, filmed in a studio and streamed to players. The randomness comes from the wheel itself, the same way it does on a casino floor. The wheel reads the winning pocket electronically, and the result on screen can be checked against the video of the ball sitting in the pocket.
Studios operate under the licensing conditions of every jurisdiction they stream to, and one studio table usually streams to many casinos at once. Our article on RNG vs live dealer roulette compares the two formats on fairness, speed and table limits.
Wheel bias: the cases that made casinos careful
A physical wheel can develop a fault: a worn pocket, a loose fret (the separator between pockets), a slight tilt, a rotor that runs out of true. Any of these can make some numbers come up more often than 1 in 37. That is called wheel bias, and it has a famous history.
In 1873 Joseph Jagger, an engineer from Yorkshire, paid clerks to record the results of the wheels at the Beaux-Arts casino in Monte Carlo. One wheel favoured certain numbers, and he bet on them until the casino caught on and began moving wheels around. In the early 1990s Gonzalo García-Pelayo and his family did something similar at Casino Gran Madrid, using computer analysis of recorded spins. The casino barred them. Spanish courts later found that studying a wheel’s results was not cheating.
Both cases had the same lesson for operators: track the results and maintain the hardware. A biased wheel is a casino’s problem far more than a player’s opportunity, because modern record-keeping now spots it long before an outsider could collect enough data.
A side story that did not involve bias: in 2004 three players at the Ritz Club in London won around £1.3 million, and the casino alleged they had used a laser scanner hidden in a phone to predict where the ball would land. Police investigated and brought no charges, and the players kept the money. The case is still argued about, and nobody has shown that the method works reliably under real table conditions.
How casinos look after physical wheels
Wheel makers such as Cammegh and TCS John Huxley build modern wheels with low-profile frets and shallow pockets, which make the ball bounce more and the result harder to predict. On the casino floor, the routine checks are simple: the wheel is levelled, the rotor and bearings are inspected, pockets and frets are checked for wear, and wheels may be swapped between tables. Electronic boards above the table log results, which also gives the casino a running record to test for bias.
Can a dealer aim? The idea of a “dealer signature” (a consistent spin that lands in a target section) survives in casino folklore. On a modern wheel, with the rotor spinning one way and the ball the other, and with many revolutions before the drop, it is not something a dealer can repeat on purpose. Casinos also vary dealers between tables, which removes whatever consistency might exist.
If you play at a land-based casino, the scoreboard showing the last results is there for entertainment. It cannot tell you what comes next.
RNG or live wheel: is one more random?
Players sometimes trust a physical wheel more because they can watch the ball, and sometimes trust an RNG more because no dealer is involved. Both are tested to the same end: every pocket equally likely, every spin independent.
The difference is in what can go wrong. An RNG game has no wear, no tilt and no dealer, so its risks sit in software, which is exactly what the labs examine before a game goes live and whenever it changes. A physical wheel has no code to audit at the pocket level, so its risks are mechanical, which is what levelling, inspection and result tracking cover. Neither format is “programmed to make you lose”. The house edge does that job openly, through the payouts, and it does it the same way on both.
One practical difference matters more than either: speed. An RNG game can run a spin every few seconds, while a standard live table spins about once a minute and a speed table about twice. More spins per hour means more money wagered per hour, and the edge applies to all of it.
Streaks you should expect
Some patterns look suspicious and are entirely normal. In a run of 37 spins on a single-zero wheel, roughly a third of the numbers will not appear at all, and some numbers will appear two or three times. Over 100 spins, a run of six or more of the same colour is common. A scoreboard that shows the same number twice in a row is not a fault either: the chance is 1 in 37 for any given spin, which at a busy table means it happens several times a day.
What you can check yourself
You cannot audit an RNG from the outside, and you do not need to. The practical checks are these:
- Confirm the operator’s licence for your province before you open an account.
- Read the game’s rules panel: wheel type, payouts, and any special rules such as La Partage.
- Keep your own record of what you wager. It tells you more about what the game is costing than any streak.
Our view is plain: at a registered operator, a standard roulette game is about as transparent as gambling gets, because the edge is printed in the payouts and anyone can check it. The genuine risk is an unlicensed site, where there is no lab, no standard and no regulator to complain to.