Percentile Calculator

Value at a percentile: enter your data and a percentile from 0 to 100, and the tool returns the data value at that position — for example, the 90th percentile of response times.

Percentile rank of a value: enter your data and one value, and the tool reports what percentage of the way up the distribution that value sits — for example, where a score of 35 falls among all scores.

There is no single agreed definition of a percentile — statisticians catalog nine common variants, labeled R-1 through R-9. This calculator uses R-7, the linear-interpolation convention that Excel's PERCENTILE.INC, Google Sheets, and NumPy's default all share, so results here match the tools most people cross-check against.

In words: the target rank h usually falls between two sorted data points, and the answer is taken proportionally between them. Rank mode runs the same formula in reverse (Excel's PERCENTRANK.INC does the same), which means the two modes round-trip cleanly: the 40th-percentile value of a data set has a percentile rank of exactly 40%. Values below the minimum or above the maximum are reported as 0% and 100% respectively.

The two directions are easy to mix up because everyday speech uses one word for both. A percentile is a data value: the 90th percentile of household incomes is a dollar amount. A percentile rank is a percentage attached to a value: an income of $85,000 might have a percentile rank of 74%. One is measured in the units of your data, the other always lives on a 0–100 scale.

Under the R-7 convention the 50th percentile equals the ordinary median, and the 0th and 100th percentiles equal the minimum and maximum. The 25th and 75th percentiles are one common definition of the quartiles — though our quartile calculator deliberately uses the median-split convention taught in most statistics courses, so its Q1 and Q3 can differ slightly from the interpolated values here. Both are correct; they simply follow different published definitions.

Percentiles dominate reporting whenever raw values mean little without context.

• Pediatric growth charts: a toddler at the 30th percentile for weight is lighter than 70% of peers — a position, not a diagnosis.
• Standardized testing: scores are reported as percentile ranks so results stay comparable across different test forms and years.
• Software performance: engineers track p95 and p99 latency because averages hide the slow requests that users actually notice.
• Salary benchmarks: compensation bands are often set between the 25th and 75th percentile of market pay for a role.

Percentiles from data are empirical; if instead you know a measurement's z-score in a normal distribution, the z-score calculator converts it to a theoretical percentile — a z of 1.28, for instance, sits near the 90th. And when all you need are the three quartile cut points plus the extremes, the five number summary calculator packages them in one pass.

Take the classic data set 15, 20, 35, 40, 50 (n = 5, already sorted). Three lookups show how the interpolation behaves:

1. 40th percentile: h = (5 − 1) × 40 ÷ 100 = 1.6, so the target sits 0.6 of the way from the 2nd sorted value (20) to the 3rd (35). Value = 20 + 0.6 × (35 − 20) = 20 + 9 = 29.
2. 90th percentile: h = 4 × 0.9 = 3.6, which is 0.6 of the way from 40 to 50. Value = 40 + 0.6 × 10 = 46.
3. Percentile rank of 35: 35 is the 3rd sorted value, at zero-based index 2, so its rank is 2 ÷ (5 − 1) = 0.5 = 50% — it is also the median of this data set.

The round trip works too: asking for the rank of 29 returns exactly 40%, undoing the first lookup. Note what a nearest-rank calculator would say instead: it can only return values that exist in the data, so its 40th percentile of this set is 20, not 29. Neither answer is wrong — they follow different definitions, which is why stating your method matters.