Sample Size Calculator
Required sample size for a margin of error.
How accurate a survey result is, given its sample size.
50% gives the widest margin, which is why pollsters quote it.
Optional. Narrows the margin when the sample is a large share of the whole group.
| Sample size | Margin of error | Range around the result |
|---|
Margin of Error Calculator works out how precise a survey result is, given how many people were asked. It is the critical value times the standard error of the proportion, and it is why a poll reporting 52 per cent with a three point margin cannot distinguish that from 49 per cent.
The result percentage affects the margin, and fifty per cent is the worst case. That is why pollsters quote the margin at fifty: it is the largest it can be for that sample size, so no individual question in the survey has a wider one. A result of ninety per cent has a substantially narrower margin on the same sample.
The sample-size table shows the brutal arithmetic of precision. Halving the margin of error takes four times the sample, because the margin falls with the square root of n. Going from a thousand respondents to two thousand improves the margin by about thirty per cent, which is why national polls cluster around a thousand: beyond that, the cost rises much faster than the accuracy.
The page shows its working, not just its answer. Where there is a formula it is printed with your own numbers substituted in, so you can check the result rather than take it on trust -- which is the difference between a calculator you can use for something that matters and one you cannot.
Everything runs inside your browser. What you type is never uploaded, never logged and never stored, and the calculator keeps working if your connection drops after the page has loaded.
Looking for something this does not cover? FreeToolsHub.org is another hub of free online tools that need no account, and it is worth a look if margin of error calculator is only part of what you are working on.
The margin of error calculator is used by writers, developers, students, marketers and anyone else who needs the job done once without installing software. Common cases include:
Because that is where it is widest. Quoting the worst case means no question in the survey has a larger margin than the headline figure.
No. It covers random sampling error only. A leading question or a sample that misses part of the population introduces error this cannot measure.
Sampling 500 people out of 2,000 tells you more than 500 out of millions. The correction narrows the margin when the sample is a large share of the population.
If the margin of error calculator is not quite what you need, these other free tools solve closely related problems, and there are 13 more in statistics calculators.
Required sample size for a margin of error.
Confidence intervals for a mean or a proportion.
Z-score with percentile and tail probabilities.
Conversion rate with uplift and revenue impact.
Three calculators: percent of a number, what percent one number is, and change.
Checked and rebuilt on 2026-10-14. The method this calculator uses is described in the section above, and corrections go through the contact page.