Educational game-integrity guide

Fair Dice, RNGs and Game Integrity Explained

Are online dice games fair? There is no universal answer. Fairness cannot be inferred from the words online, random, tested, licensed, or provably fair alone. Each term addresses a different question, and the supporting evidence must apply to the exact game, product version, rules, and jurisdiction being evaluated.

A useful review separates five questions: How are outcomes generated? Was the implementation tested or independently verifiable? What does the paytable return on average? Who regulates the activity? Does the cited evidence apply to this particular game?

This page explains those questions for educational purposes. It does not endorse, test, certify, or rank any operator, laboratory, software provider, or game.

Editorial role

Dice Rules & Odds editorial team

Research method

Desk research and reproducible probability principles; no hands-on system testing

Review status

Checked against the supplied technical standard and supporting research

Last modified

October 15, 2026

Principal sources checked

GLI-19 Standards for Interactive Gaming Systems, version 2.0; supplied mathematical and editorial research

Next step: Compare the evidence provided by physical dice, random number generator testing, and provably fair verification, then use the fairness-claim checklist.

On this page

Essential game-integrity terms

These definitions separate the technical process that generates an outcome from the mathematics that determines the value of a wager.

Random number generator (RNG)

A component used to produce game outcomes in a digital system. “RNG” is a broad description; it does not by itself identify the algorithm, source of randomness, implementation, testing status, or paytable.

Pseudorandom number generator (PRNG)

A deterministic algorithm that produces a sequence intended to appear random from an initial state or seed. Deterministic does not automatically mean predictable to a player, but implementation quality must be evaluated rather than assumed.

Provably fair

A protocol-dependent verification approach, often using cryptographic commitments and seeds, intended to let users check that an outcome was not changed after commitment. It does not establish favorable payouts, legal operation, or future winnings.

Independent roll

A roll whose probability distribution is unaffected by previous rolls, assuming fair dice and unchanged conditions.

Paytable

The documented schedule of payouts for winning outcomes. A house-edge calculation is incomplete without a defined wager, ruleset, and paytable.

Expected value

The probability-weighted average net result of all possible outcomes. It describes a long-run mathematical average, not what must happen in one play.

House edge

The expected player loss as a percentage of the specified initial amount wagered under defined rules and payouts. It is not the probability of losing a single round.

Core distinction: Randomness describes how outcomes are generated. Expected value and house edge depend on the probabilities, rules, and payouts assigned to those outcomes.

Game integrity has several separate layers

A single badge, test, license, or verification result should not be treated as proof of every kind of fairness. Evidence at one layer does not automatically establish a claim at another.

Example: random does not mean favorable

Suppose a game selects outcomes randomly and every outcome has the intended probability. The operator can still assign payouts that return less, on average, than the initial amount wagered. In that case, outcome generation may behave as specified while the wager still has negative player expected value.

To evaluate the game completely, you would need evidence about the outcome mechanism, implementation, test scope, wager rules, paytable, licensing, and jurisdiction. No single layer answers all of those questions.

What fair physical dice mean in probability calculations

Theoretical dice tables normally assume fair dice, independent rolls, and unchanged conditions. For a fair six-sided die, each face has the same theoretical probability on each roll.

Why two dice have 36 ordered outcomes

The first die has six possible results and the second die has six. Keeping the dice distinct gives:

6 × 6 = 6² = 36 ordered outcomes.

For example, first-die 2 with second-die 5 is a different ordered outcome from first-die 5 with second-die 2, even though both have a total of 7.

Why three dice have 216 ordered outcomes

Each of three distinguishable dice has six possible faces:

6 × 6 × 6 = 6³ = 216 ordered outcomes.

The table shows the mathematical sample spaces used by many dice calculations. It does not prove that a particular venue’s equipment or operating process meets applicable requirements.

Assumptions are not evidence about a specific table

A probability model can assume fair dice without establishing that a particular physical game uses suitable equipment or operating controls. Equipment specifications, inspections, handling procedures, and table rules may be governed by venue-specific or jurisdiction-specific requirements. Those details should be checked against the applicable official rules rather than treated as universal.

If the dice are fair and conditions remain unchanged, earlier rolls do not alter the probability distribution of a later independent roll. For a fuller treatment, see how ordered outcomes and independent dice probabilities are calculated.

How RNGs and PRNGs produce digital game outcomes

A random number generator is the broader outcome-generation component used by a digital game. A pseudorandom number generator is one type of mechanism: a deterministic algorithm produces a sequence from an initial state or seed.

The term “deterministic” describes the algorithmic process. It does not establish that a player can predict the sequence, nor does it establish that the implementation is secure or correct. Those questions require evidence about the actual system.

Conceptual RNG process

1. Input or state: The system begins with an internal state or other implementation-specific input.
2. Generation process: The RNG or PRNG produces a value according to its implemented method.
3. Game mapping: The system maps that value to a defined game outcome, such as a die face or dice total.
4. Rule application: The game applies its rules and paytable to determine the result of the wager.
5. Record or display: The outcome and resulting transaction are presented or stored according to the system design.

Different systems may use different methods, mappings, controls, and records. A general description of PRNGs should not be treated as evidence about a specific game, and knowing the broad concept does not provide a method for predicting or manipulating future outcomes.

What casino RNG testing can evaluate

Gaming laboratory standards can include technical, statistical, and system-security requirements for random number generators and interactive gaming systems. The supplied technical source for this page is GLI-19 Standards for Interactive Gaming Systems, version 2.0.

A published standard describes requirements. It is not, by itself, evidence that a particular product was tested, passed, approved, or licensed under that standard.

Testing may address several categories

• Implementation: Whether the tested system performs the defined outcome-generation process within the assessed scope.
• Statistical behavior: Whether observed properties meet the relevant requirements and test criteria.
• Outcome mapping: Whether generated values are translated into game outcomes as specified.
• System controls and security: Whether relevant technical protections and operational controls meet the standard within the test scope.
• Records and reporting: Whether required game or transaction information is handled as specified.

What to require before repeating a testing claim

• The identity of the testing laboratory or authority
• The standard and version used
• The exact product, game, component, and software version covered
• The scope and limitations of the assessment
• The test or report date
• Evidence that the claim applies to the currently presented implementation

Passing technical or statistical testing does not determine whether a paytable offers favorable expected value. It also cannot guarantee that a player will win a particular round or profit over time.

Read GLI-19 Standards for Interactive Gaming Systems, version 2.0 (external PDF).

What provably fair dice verification does—and does not—show

“Provably fair” generally describes a protocol intended to let a user check that an outcome was not changed after a prior commitment. Implementations often use cryptographic commitments and inputs called seeds, but the exact algorithm and verification steps are protocol-dependent.

Protocol-neutral sequence

1. Commitment: Before the outcome is finalized, the system publishes or provides a cryptographic commitment related to hidden input.
2. Inputs: The documented protocol combines the relevant server, user, sequence, or other defined inputs.
3. Outcome: The protocol generates and displays the game result according to its documented mapping rules.
4. Reveal: The information needed to check the earlier commitment becomes available as specified.
5. Verification: The user follows the exact documented algorithm to compare the commitment, inputs, and result.

Do not apply a generic verifier or an example from another service to a specific implementation. Use the exact protocol documentation, algorithm, mapping rules, and instructions for the game being assessed.

Physical dice, RNG testing, and provably fair systems compared

Conclusion: Physical controls, implementation evidence, laboratory testing, and provably fair verification can address different integrity questions. None replaces a separate review of the rules, paytable, license, legal jurisdiction, and risks of gambling.

A random game can still have a house edge

Hit rate

How frequently a wager produces a winning result.

Expected value

The probability-weighted average net result across all possible outcomes.

House edge

The expected player loss as a percentage of the specified initial wager under the stated rules and payouts.

A wager can win frequently but pay too little when it wins to offset its losing outcomes. A high hit rate therefore does not establish favorable expected value. Likewise, a properly random outcome mechanism cannot turn an unfavorable paytable into a favorable one.

To inspect commissions, pushes, exceptions, net winnings, and total-return wording, use the guide to audit dice-game rules and paytables separately from randomness.

Previous results do not make the next independent result due

The belief that independent outcomes must quickly balance is called the gambler’s fallacy. Long-term theoretical proportions do not create a schedule that individual results must follow.

Statistical testing can evaluate defined properties of an implementation or sample within a specified assessment. That is not the same as predicting a player’s next result. A short unusual sequence does not, by itself, establish bias, and successful RNG testing cannot guarantee what will happen on an individual future play.

See the full guide to independent rolls, ordered outcomes, and theoretical dice probability.

How to evaluate a dice-game fairness claim

Use this checklist before relying on language such as “fair,” “random,” “tested,” “certified,” or “provably fair.” It can be printed with the rest of this page.

1. Identify the exact game

   • Record the game and product name.
   • Identify the software or physical-table version where available.
   • Distinguish physical-table, live-dealer, and fully digital play.

2. Record the ruleset and paytable

   • Find the complete wager rules, payout schedule, exceptions, pushes, and fees.
   • Check whether quoted payouts are net winnings or total returns.
   • Do not accept a house-edge claim without a named wager, ruleset, paytable, and calculation denominator.

3. Identify the outcome-generation method

   • Look for a specific description of physical dice, live-dealer equipment, RNG, PRNG, or a documented verification protocol.
   • Treat vague statements such as “advanced randomness” as incomplete evidence.

4. Examine testing claims

   • Identify the laboratory or authority.
   • Record the technical standard and version.
   • Confirm the product and software version covered.
   • Check the test scope, date, limitations, and whether the evidence remains applicable.

5. Examine provably fair claims

   • Find the exact protocol and algorithm documentation.
   • Identify the commitment, required inputs, mapping rules, reveal process, and verification instructions.
   • Confirm that the available evidence lets a user reproduce the stated check.

6. Analyze payout value separately

   • Use the paytable and outcome probabilities to calculate expected value.
   • Do not infer favorable value from random outcomes, testing, or a successful cryptographic check.
   • Consult the paytable-audit guide for the separate calculation process.

7. Verify licensing and legal authority

   • Identify the state, tribal, commercial, land-based, or online jurisdiction involved.
   • Use the applicable official regulator’s records rather than relying on a technical badge.
   • Follow the U.S. license and legal-authority verification workflow.

8. Watch for unsupported claims

   • Unverifiable badges or seals
   • Missing laboratory, standard, product, scope, or date
   • Certification language that does not identify what was certified
   • Claims that testing or verification guarantees profit
   • Claims that one technical result establishes legality everywhere

Integrity evidence does not answer every format or legal question

A physical-table game, a live-dealer stream, and a fully digital RNG game present different evidence and dependencies. A physical game may focus attention on equipment and operating procedures. A live-dealer format adds video, communication, and remote-system considerations. An RNG game depends on digital generation, implementation, mapping, and system controls.

These delivery differences do not determine legality. Commercial, tribal, land-based, and online gaming operate under different legal and regulatory frameworks. A game’s appearance in a global catalog or foreign rulebook does not establish that it is available or lawful in a particular U.S. jurisdiction.

Fair outcomes do not make gambling risk-free

Standards, assumptions, and sources checked

This page distinguishes sourced definitions and technical-standard statements from editorial explanations. It does not report hands-on testing and does not certify any game, operator, RNG, or provably fair implementation.

Research status

Desk research only

Editorial role

Dice Rules & Odds editorial team

Last checked

October 15, 2026

Primary technical source

GLI-19 Standards for Interactive Gaming Systems, version 2.0

Source type

Gaming technical standard

Source URL

GLI-19 version 2.0 (external PDF)

Technical scope used here

RNG, system-security, and testing requirements relevant to interactive gaming systems

Mathematical assumptions

Fair six-sided dice, independent outcomes, and unchanged conditions when explaining the 36- and 216-outcome sample spaces

Editorial explanations

The integrity-layer framework, comparison table, and claim-evaluation checklist organize the sourced concepts without evaluating a specific product.

Physical-equipment boundary

Detailed manufacturing tolerances, inspection practices, and jurisdiction-specific table procedures are omitted because they require additional authoritative sourcing.

Review boundaries

• No game, system, software version, operator, or laboratory report was independently tested for this article.
• No testing badge, certification, regulatory approval, or license is implied.
• No specific provably fair result has been reproduced or presented as an audit.
• No payout or house-edge conclusion should be drawn without the exact rules and paytable.

Correction history

No meaningful corrections are recorded in the supplied publication materials for this version.

To question a definition, source, technical statement, accessibility issue, or broken link, use the editorial contact route. Please include the page title, affected passage, supporting source, and date accessed; do not send sensitive personal or gambling-account information.

Review the source hierarchy, disclosure standards, and correction process.

Continue checking probability, payouts, formats, and legal authority

• Dice Probabilities, Expected Value and House Edge

  Understand ordered outcomes, independent rolls, hit rate, and the mathematics behind expected value.

• How to Read Dice-Game Paytables

  Check payouts, commissions, pushes, exceptions, and return wording separately from outcome randomness.

• Online vs. Land-Based Dice Games

  Compare physical-table, live-dealer, and RNG delivery formats without treating one format as universally preferable.

• U.S. Gambling Laws: How to Verify Dice-Game Legality

  Locate the applicable official authority and check licensing, activity type, minimum age, and jurisdiction.

• Responsible Gambling Help and Support Resources

  Access current confidential support, practical guidance, and state-specific pathways.