A pseudo-random number generator (PRNG) is a deterministic algorithm: given the same seed, it produces the same sequence. A true random number generator (TRNG) extracts entropy from a physical process such as thermal noise, atmospheric noise, or quantum effects, so its output cannot be reproduced even with full knowledge of past values. Most software, including this tool, uses a cryptographic PRNG (CSPRNG) seeded from the operating system's entropy pool, which is fed by hardware sources and is unpredictable in practice.

It is not a TRNG. It calls the browser's Web Crypto API (crypto.getRandomValues), which is a cryptographically secure PRNG seeded from operating-system entropy. The output is unpredictable for any practical purpose, including games, raffles, and unbiased selection, but a dedicated hardware RNG or services such as random.org (which samples atmospheric noise) provide genuine physical randomness if you need a TRNG.

Math.random() in JavaScript engines is a fast non-cryptographic PRNG, historically based on algorithms like xorshift128+. Its internal state can be recovered by an attacker who observes only a handful of outputs, which means past and future values become predictable. MDN explicitly warns against using Math.random() for anything security-related. Use crypto.getRandomValues for tokens, keys, passwords, and any draw that must resist tampering.

Not in this tool. crypto.getRandomValues deliberately exposes no seed, because reproducibility would defeat its security purpose. If you need a reproducible sequence (for example, regression tests or shared simulations), use a seeded PRNG such as Mulberry32 or PCG and document the seed alongside the results. For audited drawings, document the inputs, timestamp, and full output list instead of trying to replay the draw.

Yes. The generator uses rejection sampling so that each integer in [min, max] has identical probability 1 / (max - min + 1). A naive approach using modulo on a 32-bit value introduces small bias when the range does not evenly divide 2^32; rejection sampling avoids this. Over a large number of draws the histogram converges to flat, and standard chi-square tests applied to crypto.getRandomValues output show no statistically significant deviation from uniform.

For casual fair-decision use - picking a winner from a list, choosing who pays the bill, deciding between options - yes, because each number is equally likely and unpredictable. For legally regulated drawings (raffles with prizes, audited lotteries) you typically need a documented procedure, often with a third-party witness or a publicly auditable source such as random.org's signed draws or a certified hardware RNG. Record the range, count, duplicates setting, and full result list for any drawing you may need to defend later.

Random.org is a TRNG that samples atmospheric radio noise on dedicated hardware, then offers signed and certified draws useful for audits and prize promotions. This generator is a CSPRNG running entirely in your browser, with no network call, so it is faster and works offline but does not produce a third-party audit trail. For everyday randomness both are effectively unpredictable; for regulated drawings random.org's signed records are often the safer choice.