Calculate a Z-score from a raw score and reference distribution, standardize a complete dataset, reverse a known Z-score, or convert a normal percentile. See lower-tail, upper-tail and two-tailed probabilities with substituted steps.
Z-Score Calculator with Percentile and Steps
Choose a mode, enter valid numbers and calculate from unrounded values. Fractions and scientific notation are supported.
Z-score result
Normal-model probabilities are shown separately
The standardized score and interpretation will appear here.
Standard normal curve
Calculated measures
| Measure | Value | Meaning or method |
|---|
Standardized dataset preview
| Row | Value | Deviation | Z-score | Normal percentile |
|---|
Calculation steps
- Select a mode and calculate valid values.
Recent calculations
How to Use This Z-Score Calculator
- Choose Score to Z when you know a raw score, mean and standard deviation. Choose Reverse solve when one of those values is unknown.
- Use Percentile to Z to convert a lower-tail normal percentile into a Z-score and matching raw score. Use Raw dataset when the calculator should find the mean and standard deviation first.
- Enter plain numbers. Decimals, signed numbers, scientific notation and simple fractions such as 3/4 are accepted.
- For dataset mode, decide whether your values describe the complete population or a sample from a larger population.
- Select display precision and graph shading, then press Calculate. Review the result table, normal curve and substituted steps.
The calculator keeps full available JavaScript precision during the calculation. Display precision changes only the visible rounding. It does not round your inputs or intermediate results. Dataset mode supports up to 50,000 observations and calculates from every valid entry, while the on-page row preview is capped to keep large results responsive.
Your entries remain in the current browser page. The parser checks complete tokens and rejects text, malformed fractions, nonfinite values and executable input. If you type 12kg, it reports an error instead of silently treating the entry as 12.
What Is a Z-Score?
A Z-score, also called a standard score, tells you how far a value lies from a reference mean in standard-deviation units. A positive Z-score is above the mean. A negative Z-score is below the mean. A Z-score of zero equals the mean.
Standardization helps compare values measured on different scales. For example, exam scores from two tests can be compared relative to each test's own mean and spread. A larger Z-score represents a higher relative position when the reference groups and measurement quality are comparable.
Calculating a Z-score does not require the original data to be normally distributed. Converting that Z-score into the normal percentile and tail probabilities does require a normal reference model. The calculator labels those values as normal-model results so they are not confused with an empirical percentile calculated by ranking observed data.
Z-Score Formula
For a population value x with population mean μ and population standard deviation σ, the standard score is:
For a sample described by sample mean x̄ and sample standard deviation s, the descriptive formula is:
The subtraction finds the signed distance from the mean. Division by the positive standard deviation expresses that distance in standardized units. The formula is undefined when standard deviation is zero because division by zero has no valid Z-score interpretation. A constant dataset therefore cannot be standardized with this method.
Do not confuse a descriptive sample Z-score with a hypothesis-test statistic for a sample mean. A test statistic often uses a standard error, such as σ / √n, in its denominator. This page standardizes an individual score or dataset value. It does not perform a Z-test.
Worked Z-Score Example
Suppose a score is 85, the reference mean is 80 and the standard deviation is 6. First find the signed difference: 85 − 80 = 5. Then divide by 6.
The score is about 0.8333 standard deviations above the mean. Under a standard normal model, the lower-tail cumulative probability is about 0.797672, so the score is near the 79.77th normal percentile. The upper-tail probability is about 0.202328.
| Input or result | Value | Interpretation |
|---|---|---|
| Raw score | 85 | Observed value being standardized |
| Mean | 80 | Center of the reference distribution |
| Standard deviation | 6 | Reference spread |
| Z-score | 0.833333 | 0.833333 standard deviations above the mean |
The probability values are calculated from the unrounded Z-score. If you manually look up 0.83 in a printed Z-table, your answer may differ slightly because the table starts from a rounded value.
Convert a Z-Score to a Percentile
The normal percentile is the cumulative area to the left of a Z-score under the standard normal curve. It is written as Φ(z). The standard normal distribution has mean 0 and standard deviation 1.
Upper tail = 1 − Φ(z)
Two-tailed probability = 2Φ(−|z|)
For Z = 1.5, the lower-tail probability is approximately 0.9331928. Multiplying by 100 places the score at about the 93.3193rd normal percentile. The upper tail is approximately 6.6807%. The two-tailed probability beyond ±1.5 is approximately 13.3614%.
The calculator evaluates both tails directly. This matters for large positive Z-scores because subtracting a tiny upper tail from 1 can erase useful digits. A displayed percentile near 100% does not mean the probability is exactly 100%.
Percentile to Z-score
Inverse mode solves z = Φ−1(p), where p is the lower-tail proportion. The 95th percentile gives Z ≈ 1.6448536. If the reference mean is 100 and standard deviation is 15, the matching raw score is 100 + 1.6448536 × 15 ≈ 124.6728.
Percentiles of exactly 0% and 100% have infinite theoretical Z-scores, so the tool requires a percentile strictly inside that interval.
Reverse a Z-Score to a Raw Score
Rearranging the formula recovers an original measurement from its Z-score:
With mean 70, standard deviation 10 and Z = 1.5, the raw score is 70 + 1.5 × 10 = 85. The reverse panel also solves for the mean with μ = x − zσ, or for standard deviation with σ = (x − μ) / z.
Solving for standard deviation needs extra validation. Standard deviation must be positive. If Z = 0 and x equals the mean, infinitely many positive standard deviations fit the equation, so the result is indeterminate. If Z = 0 while x differs from the mean, no solution exists. Inputs that imply a negative standard deviation are inconsistent and are rejected.
Z-Scores from a Raw Dataset
Dataset mode first calculates the arithmetic mean and selected standard deviation from the entered observations. It then standardizes the target and every dataset row using unrounded statistics. The target remains a comparison value and is not added to the dataset.
| Choice | Standard deviation | When to use it |
|---|---|---|
| Population | σ = √[Σ(xi − μ)² / N] | The entered values are the complete population of interest. |
| Sample | s = √[Σ(xi − x̄)² / (n − 1)] | The values are a sample used to describe a larger population. |
For 2, 4, 4, 4, 5, 5, 7 and 9, the mean is 5. Population standard deviation is 2, so the population Z-score for 9 is (9 − 5) / 2 = 2. Sample standard deviation is approximately 2.13809, so the sample Z-score for the same target is approximately 1.87083.
The calculator uses scaled centered deviations instead of subtracting two large squared quantities. This improves stability for datasets with a large offset and small spread. Browser numbers still have roughly 15 to 16 significant decimal digits, so differences smaller than the stored input precision cannot be recovered.
How to Interpret Z-Scores
| Z-score range | Distance from mean | Careful interpretation |
|---|---|---|
| Z = 0 | At the mean | Equal to the reference center |
| |Z| < 1 | Less than one SD | Relatively close to the mean |
| 1 ≤ |Z| < 2 | One to two SDs | Farther from the center |
| 2 ≤ |Z| < 3 | Two to three SDs | Often called unusual under a normal model |
| |Z| ≥ 3 | At least three SDs | Far into a normal-model tail |
These bands are descriptive, not universal decision rules. A value with |Z| above 2 is not automatically an error or outlier. Skewed distributions, heavy tails, small samples and mixed populations can produce large standardized values more often than a normal model predicts. Review the data-generating process before excluding an observation.
The sign shows direction, while the absolute value shows distance. A Z-score of -2 and a Z-score of 2 are equally far from the mean. They have different one-sided percentiles but the same symmetric two-tailed normal probability.
Z-Score, Percentile and T-Score Comparison
| Measure | Scale | Main use |
|---|---|---|
| Z-score | Mean 0, SD 1 after matching standardization | Express distance from a reference mean |
| Normal percentile | 0% to 100% | Express cumulative position under a normal model |
| Empirical percentile rank | Based on observed ordering | Describe rank within actual sample data |
| Scaled T-score | Often 50 + 10Z | Avoid negative or decimal standardized scores in reporting |
| Student's t statistic | t distribution with degrees of freedom | Inference when a standard error is estimated |
A scaled T-score used in educational or psychological reporting is not the same as Student's t test statistic. Always check the definition supplied with a report. This calculator reports Z-scores and normal-model probabilities only.
Common Z-Score Mistakes
- Using zero or negative standard deviation: A valid standard deviation must be greater than zero.
- Choosing the wrong dataset formula: Decide whether the values form a complete population or a sample before calculating.
- Rounding too early: Keep the mean and standard deviation unrounded until the final display.
- Calling a normal percentile empirical: Φ(z) is a model-based area. Actual ranked data can produce a different percentile.
- Treating every large |Z| as bad data: Investigate context and distribution shape before removing a valid observation.
- Confusing an individual score with a sample mean: Sample-mean inference often uses standard error rather than the raw-score standard deviation.
Calculation Method and Review Sources
The page uses the standard formulas z = (x − mean) / SD and x = mean + z × SD. Normal probabilities are evaluated numerically from the standard normal distribution. Inverse percentiles use a rational approximation and calculate from the unrounded probability.
The definitions, formula direction and probability interpretation were reviewed against the NIST normal distribution reference, the NIST standard normal cumulative table and the OpenStax standardized scores guide. The calculator distinguishes a descriptive Z-score from a statistical test and labels all percentile outputs as normal-model results.
Related Calculators
Frequently Asked Questions
What is a Z-score?
A Z-score is a unit-free standard score that states how many standard deviations a value lies above or below a reference mean.
How do you calculate a Z-score?
Subtract the reference mean from the raw score, then divide the signed difference by the positive standard deviation: z = (x - mean) / SD.
What do positive, negative and zero Z-scores mean?
A positive Z-score is above the mean, a negative Z-score is below the mean and a zero Z-score is exactly at the mean.
What is the difference between population and sample Z-scores?
A population Z-score uses the population mean and population standard deviation. A sample descriptive Z-score uses the sample mean and sample standard deviation.
How do I convert a Z-score to a percentile?
Evaluate the standard normal cumulative distribution at Z and multiply by 100. The result is a normal-model percentile, not an empirical rank.
Does calculating a Z-score require normal data?
No. Standardization itself does not require normality. Interpreting the Z-score with standard normal percentiles or tail probabilities does require a suitable normal model.
What does a Z-score of 1.96 mean?
It is 1.96 standard deviations above the mean. Under a standard normal model, about 97.5002% lies to its left and about 4.9996% lies beyond plus or minus 1.96.
Is a Z-score above 2 or 3 always an outlier?
No. Those values are far from the mean under a normal model, but an outlier decision depends on distribution shape, sample size, measurement quality and context.
What happens when standard deviation is zero?
The Z-score is undefined because the formula would divide by zero. A zero standard deviation means all positively counted dataset values are identical.
What is the difference between a Z-score and a Z-test?
A Z-score standardizes a value. A Z-test is an inferential procedure with a null hypothesis and standard error. This calculator does not perform hypothesis testing.
Educational and financial disclaimer: This calculator provides numerical estimates for learning and general analysis. Verify important results with a qualified professional or approved statistical software before using them for academic, medical, legal, employment or financial decisions.