Is an Online Random Number Generator Truly Random? How It Works

The short answer

No computer program produces "true" randomness in the sense of a coin toss governed by physics. What a good online generator does is use a cryptographically strong pseudo-random number generator: an algorithm seeded with unpredictable data collected by your operating system. The Random Number Generator on Lazyblink uses the browser's crypto.getRandomValues for this. Its results cannot be predicted in practice, which makes it fair for raffles, classroom picks, games and test data.

True random vs pseudo-random

Type	How it works	Example
True random	Measures an unpredictable physical process	Electronic noise, radioactive decay, dice
Pseudo-random (basic)	A formula produces a sequence that looks random from a starting seed	JavaScript Math.random, spreadsheet RAND
Pseudo-random (cryptographic)	A stronger algorithm seeded with entropy gathered by the operating system, designed so outputs cannot be predicted	crypto.getRandomValues, operating system random devices

MDN's documentation explains that browsers implementing getRandomValues do not use a truly random generator for performance reasons, but a pseudo-random algorithm "seeded with a value with enough entropy", and that it is "suitable for cryptographic purposes". MDN describes Math.random as "a non-cryptographic source of random numbers".

For a lucky draw, the important property is not where the randomness comes from but whether anyone could predict or influence the result. A cryptographic generator is designed so they cannot.

How the Lazyblink generator picks a number

It asks the browser for a batch of random 32-bit numbers from crypto.getRandomValues.

For each one it calculates min + (value mod range), where range is max − min + 1.

If No duplicate numbers is ticked, it skips any number already chosen.

It shows the result, or a list if you asked for more than one.

Is every number equally likely?

Almost exactly. Mapping a 32-bit number into a range with "mod" gives a tiny advantage to some values when the range does not divide 2^32 evenly. For 1 to 100, 96 of the numbers have one extra chance in about 42.9 million. In relative terms that is a difference of roughly 0.0000023%, which no raffle or game could ever detect. Security code that needs perfectly uniform output uses a technique called rejection sampling, but for picking winners this effect is negligible.

Using the tool

Open the Random Number Generator.

Set Minimum and Maximum. Both are included, so 1 and 100 can both be drawn. Make sure the minimum is smaller than the maximum.

Move the How many numbers? slider (1 to 100).

Tick No duplicate numbers if each number should appear only once.

Click Generate Random Numbers. Use Copy all to copy the results, one per line, or Regenerate for a new set.

Quick presets set the range for a dice roll (1–6), 1–100, 1–1000 and a coin flip (0–1, so you decide in advance which of 0 and 1 means heads).

If you tick No duplicate numbers and ask for more numbers than the range contains, for example 10 unique numbers between 1 and 6, you will get fewer than you asked for, because only six distinct values exist. Check the count before you announce results.

Why random results often look "not random"

People expect randomness to look evenly spread. Real random sequences are lumpier than that:

• Repeats happen. Rolling 1–6 ten times without repeats would be unusual. Getting the same number twice in a row is normal.
• Streaks happen. Five tosses of heads in a row has a 1 in 32 chance, so it will turn up regularly over many sessions.
• Clusters happen. In a draw of 10 numbers from 1 to 100, it is common to see several numbers close together.

If a result looks suspicious, that alone is not evidence of a broken generator. What matters is the method, and that nobody could choose the outcome.

Making a draw fair and transparent

For an office raffle, a giveaway or a classroom pick, the generator is the easy part. Fairness also depends on the process:

Number the entries first and publish or share the list before the draw.

Fix the rules in advance: how many winners, whether one person can win twice, and what happens if a winner does not respond.

Set the range to match the list exactly, such as 1 to 87 for 87 entries.

Draw in front of people or record your screen, and take a screenshot of the result.

Draw all winners in one batch with No duplicate numbers ticked, instead of regenerating until you like the answer.

Our guide to raffle draws without repeats walks through a full example.

For official lotteries, regulated prize competitions or anything with legal requirements, follow the rules that apply to you. These often require certified equipment or an independent witness, which a browser tool cannot provide.

When not to use a random number tool

• Passwords. A random number is not a password. Use the Password Generator, which builds a password from a character set.
• Encryption keys. Use the key-generation functions of a proper cryptographic library. MDN itself recommends generateKey() rather than raw random values for keys.
• Shuffling a long list. You can draw unique numbers and use them as positions, but a dedicated shuffle is simpler for large lists.

Privacy

The numbers are generated in your browser. Nothing you enter is sent to a server, and the results are not stored after you close the page, so copy or screenshot them if you need a record.

Summary

• Online generators are pseudo-random, not physically random.
• A cryptographic source such as crypto.getRandomValues is unpredictable in practice.
• Lazyblink uses crypto.getRandomValues; the tiny bias from range mapping is far too small to matter for draws.
• Repeats and streaks are a normal part of randomness.
• A fair draw also needs a published list, fixed rules and a recorded result.