Visual Tools
Calculators
Tables
Mathematical Keyboard
Converters
Other Tools

How to Find a Normal Percentile

A percentile question gives an area and asks for the value that cuts it off. Turn the area into a left-tail probability pp, find the z-score with Φ(z)=p\Phi(z) = p, and undo the standardization: x=μ+zσx = \mu + z\sigma.

Recognize the task

The question gives a percentage and asks for a value: "the score needed for the top 10%10\%", "the cutoff for the lowest 5%5\%", "the middle 95%95\% of heights", "set the mean so that only 2%2\% are underweight". Test: the unknown is xx, μ\mu or σ\sigma, not a probability.
Quantiles of the normal are explained in the normal quantiles lesson section.

Steps

StepOn scores N(70,102)N(70, 10^2), the cutoff for the top 10%10\%
1Turn the wording into a left-tail area pp. "Top 10%10\%" leaves 90%90\% to the left, so p=0.90p = 0.90. "Bottom 5%5\%" is p=0.05p = 0.05. "Middle 95%95\%" puts 2.5%2.5\% in each tail: p=0.975p = 0.975 for the upper end.p=0.90p = 0.90
2Find zz with Φ(z)=p\Phi(z) = p. Use the inverse normal in software, or find the closest area inside a table. Lower percentiles give negative zz.z0.90≈1.2816z_{0.90} \approx 1.2816
3Undo the standardization. From z=(x−μ)/σz = (x - \mu)/\sigma (standardization): x=μ+zσx = \mu + z\sigma.x=70+1.2816⋅10x = 70 + 1.2816 \cdot 10
4Compute and answer in units. If the unknown is μ\mu or σ\sigma instead, solve the same equation for it.x≈82.82x \approx 82.82 points

Worked examples

Example 1 · a top percentage

upper cutoff
scores N(70,102)N(70, 10^2), top 10%10\%
left area 0.900.90
z0.90≈1.2816z_{0.90} \approx 1.2816
inverse normal
x=70+1.2816⋅10≈82.82x = 70 + 1.2816 \cdot 10 \approx 82.82

Example 2 · the middle 95%

two cutoffs
scores N(70,102)N(70, 10^2), middle 95%95\%
2.5%2.5\% in each tail
z0.975≈1.96z_{0.975} \approx 1.96, z0.025≈−1.96z_{0.025} \approx -1.96
symmetric about 00
70±1.96⋅1070 \pm 1.96 \cdot 10
from 50.450.4 to 89.689.6 points

Example 3 · solving for the mean

unknown μ
fill weight N(μ,42)N(\mu, 4^2) g; only 2%2\% below 500500 g
500500 is the 22nd percentile
z0.02≈−2.0537z_{0.02} \approx -2.0537
lower tail, negative zz
500=μ−2.0537⋅4500 = \mu - 2.0537 \cdot 4
x=μ+zσx = \mu + z\sigma
μ≈508.21\mu \approx 508.21 g
Find any quantile in the continuous distributions calculator.

See it happen

A probability question fixes a point on the axis and asks for the shaded area; a percentile question fixes the area and slides the point until the shading matches. Because every normal is the standard one stretched by σ\sigma and shifted by μ\mu, the point on the standard curve, zz, maps to μ+zσ\mu + z\sigma on any other. Slide the cutoff in the normal distribution visualizer.

Common mistakes

Using the right-tail area

✗ top 10%10\%: find zz with Φ(z)=0.10\Phi(z) = 0.10, z=−1.2816z = -1.2816
✓ the area to the left is 0.900.90: z=1.2816z = 1.2816

Middle 95% with p = 0.95

✗ z0.95≈1.6449z_{0.95} \approx 1.6449, so 70±16.4570 \pm 16.45
✓ each tail holds 2.5%2.5\%: z0.975≈1.96z_{0.975} \approx 1.96, so 70±19.670 \pm 19.6

Dropping the sign of a lower z

✗ bottom 2%2\%: 500=μ+2.0537⋅4500 = \mu + 2.0537 \cdot 4, μ≈491.79\mu \approx 491.79
✓ the lower cutoff is below the mean: z=−2.0537z = -2.0537, μ≈508.21\mu \approx 508.21