What Is a Percentile? The 90th Percentile Explained

What the percentile rank actually measures

A percentile turns a raw value into a position within a distribution. To find it, you sort all the values from lowest to highest and ask what fraction sit at or below the value of interest. Multiplying that fraction by 100 gives the percentile rank.

So a percentile is inherently relative: the same raw score can be a high percentile in one group and a low percentile in another. It always describes a value in the context of a specific set of data.

Reading the 90th percentile correctly

Being at the 90th percentile does not mean you scored 90 out of 100. It means you outperformed about 90% of the group. A child at the 90th percentile for height is taller than roughly 9 out of 10 peers, regardless of their height in centimeters.

The median is just the 50th percentile — the middle value where half the data lies below and half above. Quartiles are the 25th, 50th, and 75th percentiles, dividing the data into four equal parts.

Percentile versus percentage

A percentage measures a part of a whole: scoring 80% means you got 80 of every 100 possible points. A percentile measures rank against others: the 80th percentile means you beat 80% of the group. You can score 80% on an easy test and still land in a low percentile if almost everyone else scored higher.

Because of this, percentiles are common where relative standing matters more than the raw number: standardized tests, growth charts, income distributions, and response-time targets in engineering.

Frequently asked questions

Can you be in the 100th percentile?

Usually not. By most definitions the top value cannot exceed 100% of the data lying strictly below it, so results are typically reported up to the 99th percentile rather than the 100th.

Why do calculators give slightly different percentiles?

There are several accepted methods that differ in how they interpolate between data points and whether they use inclusive or exclusive ranking. For large datasets the differences are tiny, but for small samples they can be noticeable.

Is a higher percentile always better?

It depends on what is measured. A high percentile on a test is good, but a high percentile for server response time means slow responses, which is bad. The percentile is neutral; the interpretation depends on the metric.

How many data points do you need for percentiles to be meaningful?

There is no hard rule, but extreme percentiles like the 95th or 99th need enough data to be stable. With only a handful of values, a single point can swing a high percentile dramatically.