Steve Schütz's OrganizationVisit Website →
← Back to blog

Students: Convert Percentiles to Z Scores, 95th ≈ z +1.645

October 5, 2026
Students: Convert Percentiles to Z Scores, 95th ≈ z +1.645

Percentiles rank where you stand among others, while z-scores measure how far you are from the mean in standard deviations. You can convert between them only by assuming or modeling the distribution, commonly the normal distribution, and the tool for that conversion is the standard normal cumulative distribution function or its inverse.

TL;DR:

  • Percentiles measure relative standing within a dataset, while z-scores quantify how many standard deviations a value is from the mean, assuming normality.
  • Converting between them is valid only under the assumption that the data follows a normal distribution, with 1.645 corresponding to the 95th percentile in a normal model.
  • Percentiles are less affected by outliers and provide intuitive rankings, making them more suitable for general audiences, while z-scores are better for statistical comparisons across different variables.
  • Different software may use slightly different methods to calculate percentiles, so comparisons across tools should consider these interpolation differences.
  • Reporting both percentiles and z-scores offers a fuller understanding, especially when communicating results to both technical and non-technical audiences.
Testuno
testuno.app
Explore Your Cognitive Profile
Testuno offers quick self-assessments with estimated scores, real-time benchmarking, and detailed percentile rankings across cognitive skills.
Take an assessment

Table of Contents

What Is a Z-Score and What Does It Measure?

A z-score measures how many standard deviations a data point lies from the mean of its distribution. The formula is z = (X minus μ) divided by σ, where X is the raw score, μ is the population mean, and σ is the population standard deviation. When working from a sample instead of a full population, you swap in the sample mean and sample standard deviation, though the logic of the formula stays the same.

The sign and size of a z-score both carry meaning:

  • A z-score of 0 means the value sits exactly at the mean.
  • A positive z-score means the value is above the mean; a negative z-score means it's below.
  • The magnitude tells you how many standard deviations separate the value from the mean, so a z of 2.0 is twice as far from the mean as a z of 1.0.

Z-scores connect directly to the empirical rule for roughly normal data: about 68% of values fall within one standard deviation of the mean, about 95% fall within two, and about 99.7% fall within three. That rule gives you a fast mental check. If someone reports a z-score of 3, you already know the value is unusual, far out in the thin part of the curve, before you touch a calculator.

Because z-scores strip away the original units and replace them with a universal scale centered on zero, they let you compare apples to oranges: a test score, a height measurement, and a reaction time can all be placed on the same standardized footing once converted to z.

What Is a Z-Score and What Does It Measure? — overview diagram

What a Percentile Actually Tells You

A percentile indicates the percentage of scores at or below a given value in a data set. The 50th percentile is the median: half the scores fall below it, half above. Quartiles are just percentiles in disguise, with the first quartile equal to the 25th percentile, the second quartile equal to the 50th, and the third quartile equal to the 75th.

It helps to separate two things people often blend together:

  • Percentile rank describes where a specific score sits relative to everyone else, expressed as a percentage of the group scoring at or below it.
  • Percentile value is the reverse: given a target percentage, it's the actual score that corresponds to that cutoff in the data.

Both describe the same relationship, just read in opposite directions, and mixing them up is a common source of confusion in test reports.

A less obvious wrinkle is that percentile calculations are not perfectly standardized across software. Different programs use different interpolation methods when a percentile falls between two observed data points, and those methods can produce slightly different numbers from the same raw data. If you're comparing percentile figures from two different tools, it's worth checking which calculation method each one uses before assuming the numbers mean exactly the same thing.

Rank vs Distance: When Converting Between Them Makes Sense

Percentiles and z-scores answer fundamentally different questions, and that difference is the reason converting between them isn't always straightforward. A percentile is an ordinal measure: it tells you your position in a lineup, nothing about how far apart the people in that lineup actually are. A z-score is an interval measure: it tells you distance from the mean in standardized units, which lets you do arithmetic on it in ways that make no sense with a plain rank.

Here's the sequence that makes conversion possible:

  1. Decide whether a normal distribution is a reasonable model for your data.
  2. If it is, use the standard normal cumulative distribution function to move from a z-score to a percentile, or its inverse to move from a percentile to a z-score.
  3. If it isn't, recognize that any percentile-to-z conversion you compute is only valid under the normal assumption you've imposed, not a property of your actual data.

Under a normal model, the mapping is clean. The 95th percentile corresponds to approximately z = +1.645, a figure pulled straight from the standard normal table and one of the most frequently cited anchor points in introductory statistics. That single number is doing a lot of work: it's the cutoff analysts reach for constantly when setting one-tailed significance thresholds at the 5% level.

One statistic worth sitting with: the 95th percentile sits at roughly z = +1.645 under the normal model, meaning a score just 1.645 standard deviations above the mean already outperforms 95% of a normally distributed population.

Outside the normal case, things get murkier. Percentile ranks, especially when assigned to raw test data, are often approximately uniform rather than normal in their own right. That matters because, as the UCLA Stata guidance on percentile rank data points out, specialists frequently transform those uniform percentile ranks into z-scores using an inverse normal function specifically so the resulting values become usable in parametric statistical methods. That transformation is a modeling choice, not a discovery about the data's true shape.

Skewed distributions sharpen the problem. In a skewed data set, the gap between the 90th and 95th percentile in raw score terms might be nothing like the gap between the 95th and 99th, even though the corresponding z-score differences under a normal assumption would imply a smooth, even spacing. Extreme tails are where this breaks down hardest: two people both reported at the "99th percentile" on different tests might have wildly different underlying z-scores if the true distributions weren't normal to begin with.

How to Convert Between Percentiles and Z-Scores

Converting in either direction follows a short, repeatable sequence once you've accepted the normal distribution assumption.

Percentile to z-score:

  1. Take your percentile and divide by 100 to express it as a cumulative probability, so the 95th percentile becomes 0.95.
  2. Look up that cumulative probability in a standard normal table, or run it through an inverse-normal function on a calculator or in software.
  3. Read off the corresponding z-value. For 0.95, the UCLA Stata FAQ and NIST both point to the same standard result of approximately z = +1.645.

Z-score to percentile:

  1. Compute the cumulative distribution function value for your z-score, either from a table or a calculator function.
  2. Multiply that probability by 100 to express it as a percentile.
  3. Round sensibly. A percentile reported to four decimal places rarely adds useful precision for a reader.

Several common tools make this a one-line operation:

  • In Stata, the UCLA guidance gives the command generate zscore = invnorm(pctrank/100) to transform a percentile rank column directly into z-scores.
  • In Python, scipy.stats.norm.ppf(p) returns the z-score for cumulative probability p, while scipy.stats.norm.cdf(z) goes the other direction.
  • In Excel, NORM.S.INV(p) converts a probability to a z-score, and NORM.S.DIST(z, TRUE) converts a z-score back to a cumulative probability.

A printed z-table is fine for a single lookup or a classroom exercise, but any time you're converting a whole column of scores, a calculator function or a one-line script saves time and avoids transcription errors from squinting at a table.

Quick Reference: Percentile and Z-Score Anchors

A handful of percentile values come up constantly enough that memorizing their z-score equivalents saves time.

These figures come from standard normal distribution tables used across introductory statistics courses, and they're the same anchors referenced in NIST's handbook on percentile calculation.

Worked example: say a student's raw score on an exam is 118, and the exam's scores have a mean of 100 and a standard deviation of 15. The z-score is (118 minus 100) divided by 15, which equals 1.20. Looking at the standard normal table, a z of 1.20 corresponds to a cumulative probability of about 0.885, so that student sits at roughly the 88.5th percentile.

When reporting results, round z-scores to two decimal places and percentiles to the nearest whole number or tenth. Extra decimal places imply a precision that the underlying normal-model assumption usually can't support, since real score distributions only approximate normality.

When to Report a Percentile and When to Report a Z-Score

The right metric depends on who's reading the number and what they're going to do with it.

  • Percentiles work well for nontechnical audiences because "better than 90% of test takers" needs no statistical background to understand, and percentiles remain valid even when the underlying data isn't normal.
  • Z-scores work better for statistical comparisons, especially when you need to combine or compare measures across different scales, or when a parametric method downstream assumes normally distributed inputs.
  • Z-scores are preferred when researchers need to compare observations across different variables, because standardizing to a mean of 0 and a standard deviation of 1 puts everything on the same footing.

Pro Tip: When you have room to report both, do it. A percentile gives readers instant intuition, and a z-score preserves the interval-scale information a percentile alone throws away.

Documenting your assumptions matters more than picking a side. If you convert a percentile to a z-score, say explicitly that you assumed normality to do it. That one sentence protects you from readers over-interpreting a converted figure as if it were measured directly.

Mistakes That Trip Up Percentile and Z-Score Interpretation

A few errors show up again and again in how people read these numbers.

  • Reading a percentile as percent correct is the most common mistake. The Britannica entry on percentiles notes that on a harder exam, a high percentile can correspond to a surprisingly low percent-correct rate, because percentile reflects relative standing, not raw performance.
  • Different percentile algorithms produce slightly different results from identical data, so comparing percentile figures across two software packages without checking their calculation method invites small but real discrepancies.
  • Percentiles are bounded between 0 and 100, while z-scores are technically unbounded. That means two scores both sitting at the "99th percentile" can have meaningfully different z-scores once you look past the rounding, especially out in the tails where percentiles compress and lose resolution.

How Testuno Reports Percentiles Across Its Assessments

When building our testing platform, percentile benchmarking became the default way we report results, including on an IQ assessment with multiple subtests. We show where a given score falls relative to other test takers in real time, and we map raw performance into standardized comparisons that readers can act on immediately. Anyone who wants a deeper record of their results can download a full report after finishing.

Where Percentiles and Z-Scores Get Used in the Real World

Education relies on percentiles constantly, from standardized test reports to grade distributions, because parents and students grasp "75th percentile" faster than a raw score on an unfamiliar scale. Grade-profile reporting tools in education technology use the same relative-standing logic to show students where they land within a class.

Psychology and cognitive testing lean more heavily on z-scores, particularly when a test combines several subtests measured on different scales. Converting each subtest to a z-score before combining them is what makes a composite score meaningful rather than an apples-to-oranges average.

Medicine uses both, often side by side. Growth charts for children report percentiles because they're intuitive for parents, while clinical researchers analyzing lab values often convert those same measurements to z-scores to run statistical comparisons across patient groups. Standardized exams in professional training, including medical licensing exams, report percentile standing specifically because it communicates relative performance without requiring the reader to understand the underlying score distribution.

Career assessment platforms frequently report standardized scores alongside percentiles so that results on different subscales can be compared on a common footing even though each subscale might have its own raw-score range.

Which Measure Holds Up Better Against Outliers?

Percentiles are built on rank order, so a single extreme outlier barely moves anyone else's percentile. If the highest score in a group of 100 doubles, the people sitting at the 50th or 75th percentile stay exactly where they were, because their relative position among the other 99 scores hasn't changed.

Z-scores respond very differently. Because the formula depends on the mean and standard deviation of the whole data set, one extreme outlier can drag the mean and inflate the standard deviation, which then compresses everyone else's z-scores toward zero. A single wildly high score can make an otherwise strong performance look average once it's converted to a z-score, simply because the reference distribution shifted underneath it.

This is part of why percentiles are considered sturdier for reporting to a general audience working with messy real-world data, while z-scores demand a bit more caution, and ideally a check for outliers, before you trust them as a comparison tool. Neither measure is wrong; they just fail in different places, and knowing which failure mode applies to your data tells you which one to lean on.

Picturing Percentiles and Z-Scores on the Normal Curve

On a standard normal curve, z-scores sit along the horizontal axis, evenly spaced in standard-deviation units, with zero marking the peak at the mean. Percentiles correspond to the area under the curve to the left of a given point, so the 50th percentile is the z = 0 line that splits the curve exactly in half.

Normal curve linking percentiles and z-scores

That's the visual key to why the two measures diverge near the tails. Out past z = 2, the curve flattens dramatically, so a large stretch of the horizontal axis, a big change in z-score, corresponds to only a tiny sliver of additional area, meaning the percentile barely budges. Near the center, the opposite happens: a small horizontal move near z = 0 sweeps through a much larger share of the area, so percentiles change quickly for small z-score shifts right around the mean. This is exactly why extreme scores can shift noticeably in z-score terms while their percentiles look nearly frozen at "99th" or above.

What I'd Tell a Student Comparing These Two Numbers

If you're a student deciding what to report to a teacher or put on a resume, lead with the percentile since it's instantly readable, but keep the z-score on hand if anyone asks how you compare across different tests or subjects. My one-line rule: when you can, state both the rank and the standardized distance, because together they tell a fuller story than either one alone.

— Steve

See Where You Stand With Testuno's Quick Assessments

Curious how your own numbers look once they're benchmarked? Our quick IQ test gives you instant percentile results with no sign-up required, so you can see exactly where a raw score lands relative to other test takers in minutes.

Testuno

From there, our full catalog of assessments covers cognitive skills, personality, and relationship insights, each with the same real-time percentile benchmarking and a downloadable report if you want to keep the details for later.

FAQ

How do I convert percentiles to z-scores?

Divide the percentile by 100 to get a cumulative probability, then look up that probability in a standard normal table or run it through an inverse-normal function like NORM.S.INV in Excel. For example, the 95th percentile converts to a z-score of about 1.645 under the normal distribution assumption.

Which percentile is better, higher or lower?

For most positive outcomes, a higher percentile means a stronger relative position, since it places you further along the distribution compared to others. Whether "higher is better" depends entirely on what's being measured, since a high percentile on a measure like blood pressure wouldn't be a good thing.

What is the z-score for a 95% normal distribution?

The z-score corresponding to the 95th percentile under a standard normal distribution is approximately +1.645.

Is it better if the z-score is higher or lower?

A higher z-score places a value further above the mean, which is generally the favorable direction for measures where more is better, such as a test score. As with percentiles, whether higher is "better" depends on what the underlying variable actually measures.

What is the difference between a percentile rank and a raw percentile value?

A percentile rank tells you the percentage of a group scoring at or below a specific score, while a percentile value is the actual score that corresponds to a chosen percentage cutoff. They describe the same relationship read in opposite directions.

Sources