How casino games are tested

How casino games are tested: the short answer

Before an online slot, roulette wheel or blackjack table can take a single stake from a player in Britain, its software has to be examined by an independent laboratory that the Gambling Commission has approved. That lab checks the random number generator, plays the game’s maths through millions of simulated rounds, compares the result with the advertised return to player, and confirms the rules on screen match what the code does. Only when the report has been filed with the Commission can the game go live.

Testing does not stop on launch day. The business running the game has to watch how much it actually pays back, investigate unusual patterns and complaints, and have its whole change-control process audited by an approved lab every year. These duties come from the Commission’s Remote gambling and software technical standards (opens in a new tab) (the RTS) and a companion document called the testing strategy, both last updated on 31 October 2025. Complying with both is a licence condition, which is why every operator listed in our guides hub works to the same baseline.

Everything here applies to games offered under a Gambling Commission licence. A site without that licence has no duty to test anything, which is one reason to check a casino licence before you sign up.

Who tests online casino games in the UK

Three groups share the work. The game studio designs the maths and builds the software. An approved test house examines it independently. The licensed operator (or the supplier hosting the game for it) is responsible for submitting the results, keeping a games register up to date and monitoring the game once players are using it. The Commission sets the standards and can take enforcement action when they are missed, but it does not test games itself.

The Commission publishes a list of approved test houses, which as of 2 March 2026 included eCOGRA, several Gaming Laboratories International (GLI) entities, BMM Testlabs and others. Our separate explainer on casino testing labs such as eCOGRA and GLI covers who they are and how they win approval; this page follows the game itself.

Step one: testing the random number generator

Every digital casino result starts with a random number generator. RTS 7 requires results to be “acceptably random”, which the guidance explains as uniformly distributed output that matches expected probabilities and cannot practically be predicted without knowing the algorithm and its seed. A game that quietly adjusts its odds to make up for a run of payouts (the RTS calls this adaptive behaviour, or a compensated game) is banned outright.

According to the testing strategy, a lab’s RNG work includes reviewing the documentation, researching known weaknesses in the chosen algorithm, reading the source code to confirm it matches that documentation, and running statistical tests on both the raw numbers and the scaled numbers the game actually uses. Scaling matters: an RNG might produce perfect numbers between 0 and 4 billion, but if the code that turns them into a roulette pocket between 0 and 36 is biased, the wheel is not fair.

A worked example of scaling bias

Suppose a crude routine took a random number from 0 to 99 and used the remainder after dividing by 37 to pick a European roulette pocket. Numbers 0 to 25 would each come up three times in every hundred draws, while 26 to 36 would come up only twice. Over 1,000,000 spins, pocket 5 would land roughly 30,000 times and pocket 30 roughly 20,000 times, instead of about 27,027 each. Statistical tests on scaled output exist to catch exactly this kind of flaw.

Step two: checking the maths and the RTP

Once the RNG is trusted, the lab turns to the game. The testing strategy lists verification of the game design, including its maths, its artwork, the rules as shown to players and the theoretical return to player. Software testing then uses simulation, emulation and manual play. Simulation is the headline check: the lab runs the game’s logic through a very large number of rounds to verify that the actual RTP falls within an acceptable range of the expected figure.

Here is why volume matters. A slot with a theoretical RTP of 96% returns, on average, £96 for every £100 staked over its whole life. Over 100 spins at £1, though, a player could easily finish with £20 or £300, because volatility swings results around the average. Only across tens of millions of simulated spins does the measured figure settle close to 96%, which is why labs measure it at scale rather than by playing a few sessions.

UK-specific checks: speed, autoplay and honest displays

Games built for Britain have to meet design rules that labs in other markets may not test for. Since 31 October 2021, slots must take at least 2.5 seconds per game cycle, cannot offer autoplay, turbo or slam-stop, and must not celebrate a return that is equal to or smaller than the stake. From 17 January 2025 similar rules reach other online casino games, with a minimum of 5 seconds per game cycle on non-slot titles. The game design rules page has the full list.

RTS 7C adds a fairness rule players rarely hear about: games must not mislead about the likelihood of winning. A digital coin must land heads half the time, separate simulated dice must behave independently, and a game cannot swap one losing result for a more tantalising near miss. The consequences of getting this wrong are real. In June 2026 the supplier Stakelogic BV agreed to pay £122,835 after some of its slots ran faster than the 2.5-second minimum, as the Commission’s enforcement notice on Stakelogic (opens in a new tab) explains.

How a tested game is released to UK players

A passing test is not the end of the paperwork. The testing strategy says new games and RNGs can only be released once testing is complete and the report has been sent to the Commission. Reports are uploaded to the licensee’s games register through the Commission’s eServices system, and an online casino using a supplier’s games still has to keep its own register current.

1. The studio finishes the game’s maths model, paytable and artwork for the British market.
2. An approved test house reviews the documentation, tests the RNG and runs simulations of the game.
3. The lab checks the rules screen, paytable and UK design features, then issues a signed report with a certificate reference and digital signature for the exact software version.
4. The licensee uploads the report to its games register through eServices before release.
5. The game goes live in the casino lobby, and live RTP monitoring begins.

Each test report must record details such as the test house and signing supervisor, the licensee, the test date, the certificate reference, the game’s RTP, software number and digital signature, the platforms and channels covered, and any versions it replaces. That level of detail lets the Commission match the game on your screen to the version a lab actually examined.

What happens when a tested game is updated

Games change after launch: new artwork, bug fixes, a mobile layout. The testing strategy splits updates into minor ones, which do not affect fairness and can go out without external retesting, and major ones, which do and must be retested and reported before release. Licensees and labs judge which is which, guided by examples in an annex. If a platform or RNG change could affect fairness, a representative sample of games of each type has to be retested before the updated system launches.

Every change, major or minor, must be logged in change-control records available for inspection, with a personal management licence holder responsible for the process. Browser and operating system updates generally do not need external retesting, though the licensee should confirm its games still behave correctly. RTS 7D adds that a game’s rules or paytable may not be changed while it is available to play; changes are made with the game offline and flagged to customers, for example with a “new odds” notice.

Testing after launch: live RTP monitoring

A game can pass every lab test and still develop a fault in production, so the Commission expects monitoring of real play. Licensees must ensure there is enough live RTP monitoring to catch both underpayment and overpayment. Where a supplier hosts games for many casinos, the supplier usually does this because it holds the combined transaction data, and contracts must say who is responsible.

The Commission does not publish numeric tolerances; each game’s acceptable range should reflect its volatility. It does say that one measurement a month is inadequate for popular games and gives daily measurements over a rolling 30-day window as an acceptable example. Monitoring should not be so aggregated that an error affecting one platform or market is hidden. Complaints about fairness must be properly investigated, and the testing strategy says that simply telling a player a game “met testing standards” is not a sufficient explanation.

The annual games testing audit

Once a year, licensees that hold a gambling software licence alongside a remote casino, bingo or betting licence must have an approved test house audit their games process. The minimum scope covers a random sample of major and minor updates (to confirm each was classified correctly), adherence to change controls, the list of games offered to British customers, and whether effective live RTP monitoring is in place. Issues found must be reported even if they have already been fixed.

Results must be counter-signed by the relevant licence holder and submitted through eServices. Alongside this, a separate annual security audit checks the systems that generate random numbers and store game results against ISO/IEC 27001:2022-based requirements, which our guide to casino account security touches on from the player’s side.

How live dealer games are checked

Live casino tables use physical wheels, cards and dice, so the RNG work above does not apply in the same way. The RTS has a dedicated standard (RTS 17) for live dealer studios, and RTS 7 includes guidance for mechanical randomness: materials should not degrade, the physical properties of balls or cards should stay unaltered, and players must not be able to interfere with the mechanism. Game-show titles that add a digital multiplier or bonus wheel on top of a physical draw combine both kinds of randomness, and the digital part is tested like any other RNG game.

The 5-second minimum game cycle introduced on 17 January 2025 applies to non-slot RNG casino games; a live table’s pace is set by the dealer and the betting window. If you want to compare formats, our live casino explainer above sets out the differences.

What game testing means for you as a player

Testing gives you one thing of real value: confidence that the game behaves as described. It does not tilt anything in your favour. A tested slot with a 96% RTP still keeps an average of £4 from every £100 staked over its lifetime, and a tested European roulette wheel still carries a 2.7% house edge. Losses are the expected outcome over time, so set a budget before you play and treat any win as luck. The house edge explainer shows the maths.

Common myths fall apart once you know how testing works. A slot is not “due” to pay because it has been quiet, since RNG outcomes are independent (see the gambler’s fallacy). Casinos cannot switch a game to a “tight” setting at night, because the tested version is fixed and logged. And changing your stake does not change the odds per spin.

How to check a tested game yourself

RTS 3 requires the rules, a description of how winners are determined, and either the RTP or the probability of winning events to be easily available before you commit to a bet. That means you can do a quick check on any game in a licensed lobby.

1. Open the game and find the information or help screen (often an “i” or menu icon).
2. Note the RTP; some games come in several RTP versions, so read the figure in the live game rather than a review site.
3. Read the paytable and any feature rules, including the maximum win.
4. Scroll to the casino footer and note the licence holder and account number.
5. Confirm that business on the Gambling Commission’s public register (opens in a new tab) and check its remote casino licence is active.

Our guides to checking a game’s RTP and reading game help screens walk through those screens in more detail.

UK game testing rules: a dated timeline

Dates are taken from the Commission’s own publication notes and enforcement notices, checked on 20 October 2026.

Licensed casinos where tested games are the norm

Any casino holding a Gambling Commission remote casino licence must offer only games that have been through the process above. The three below are licensed brands we cover, each with its licence holder named so you can confirm it on the register yourself. Offers were last checked on 5 October 2026; you must be 18 or over and terms apply.

• Coral Casino is run by LC International Limited (account 54743). New players aged 18+ who deposit and bet £10 or more on slots get 100 free spins worth 10p each on selected games, to be accepted within 48 hours and used within 7 days, with no wagering on winnings; some payment methods such as PayPal and prepaid cards don’t qualify. See Coral’s casino page.
• PlayOJO is run by Skill On Net Limited (account 39326). New UK customers aged 18+ (excluding Northern Ireland) who make a first deposit of at least £10 receive 80 free spins on Big Bass Bonanza at 10p each, with winnings paid in cash and no wagering. More on our PlayOJO page.
• Casumo is run by Recro Limited (account 61549). New customers aged 18+ who opt in and deposit at least £20 get a 50% bonus up to £100, which must be wagered 10x within 7 days, with a £5 maximum bet while the bonus is active. Read our Casumo overview.

18+. Gambling should be entertainment that costs money, not a way to make it. Set a deposit limit before you play: how to set deposit limits. Free, confidential help: National Gambling Helpline 0808 8020 133.