A z-score (also called a standard score) measures how many standard deviations an observation is from the mean. z = (x − μ) / σ. Positive z = above mean; negative z = below; zero = at the mean. A z of 2 means the observation is 2 standard deviations above the mean — relatively unusual in a normal distribution (~2.3% of values).

What's a 'good' or 'bad' z-score?

Context-dependent. For test scores, percentile rank, or measurements where higher is better, positive z is good. For risk metrics, error rates, or where lower is better, negative z is good. As a general rule: |z| < 1 is typical (~68% of values), |z| < 2 is mildly unusual, |z| < 3 is rare, |z| ≥ 3 is very rare (<0.3% of normal data).

How does the percentile relate to z?

Percentile = the percentage of values that fall AT OR BELOW the observation. For a z-score of 1.0, the percentile is ~84.13% — 84% of normally-distributed values fall below that observation. The calculator uses the standard normal CDF (Φ(z)) to convert z to percentile.

What's the 'rarer than X% of values' line?

It tells you what fraction of observations are MORE extreme than yours in the tail direction. For a z-score of +2.5: 99.38% of values are below you; 0.62% are above. We report it as 'rarer than 0.62% of values' — meaning only 0.62% of the distribution is in the more-extreme tail. Useful for outlier interpretation.

Does z-score only work for normal distributions?

The CALCULATION (x − μ) / σ works for any distribution — it just measures how-many-SDs-from-mean. But the PERCENTILE interpretation (z = 1.0 → 84th percentile) is only correct for normal distributions. For skewed or heavy-tailed data, z = 1.0 might be at the 75th percentile or the 90th — the calculator's percentile assumes normality.

When is the value → z direction useful?

When you want to standardize observations across different scales or distributions. Convert SAT (μ=1050, σ=200) and ACT (μ=21, σ=5) to z-scores to compare student performance head-to-head. Convert blood pressure readings to z-scores against age-specific norms. Convert investment returns to z-scores against historical mean and SD. Standardization makes apples-to-apples comparison possible.

When is the z → value direction useful?

When you know a percentile and want the matching observed value. Examples: 'What height is the 95th percentile of US men?' → use μ=70 inches, σ=3, find x for z=1.645. 'What test score corresponds to the top 10%?' → use distribution parameters, find x for z=1.282. The reverse calculation is essential for setting thresholds.

What are critical z-values for common significance levels?

Two-tailed (most common): α=0.05 → ±1.960; α=0.01 → ±2.576; α=0.001 → ±3.291. One-tailed: α=0.05 → ±1.645; α=0.01 → ±2.326. Use these as decision thresholds in hypothesis testing — if |computed z| > critical z, reject the null hypothesis.

What's the empirical rule (68-95-99.7)?

For normally-distributed data: ~68.27% of values fall within ±1σ of the mean. ~95.45% within ±2σ. ~99.73% within ±3σ. Used as a quick mental check on observations: anything beyond ±3σ is rare enough that it suggests either an outlier or a non-normal distribution. The rule is approximate; the calculator's percentile is exact.

Can I have a z-score larger than 3?

Yes — z-scores have no upper or lower limit. A z of 4 is very rare in normal data (1 in 31,000); a z of 5 is essentially impossible (1 in 3.5 million) UNLESS the data is non-normal. Large z-scores almost always signal: (1) a data-entry error, (2) a non-normal distribution, or (3) a genuine outlier worth investigating.

How is the percentile calculated for negative z?

Same formula: percentile = Φ(z) × 100. For z = -1.5, percentile ≈ 6.68% — meaning the observation is below ~93% of values. The 'rarer than' line then reports it as 'rarer than 6.68% of values' — the smaller tail. Negative z is treated symmetrically with positive z; the underlying CDF is monotonic.

What if my σ is zero?

z-score is undefined when σ = 0 (all values identical; no spread). The calculator returns an error rather than dividing by zero. If you genuinely have a 'distribution' with σ = 0, there's nothing to standardize — every observation equals the mean by definition.

MathFree · No signup · 49.5K/month · $1.30 CPC

Z-Score Calculator — Bidirectional (Value ↔ z) + Percentile

Drop a value, mean, and SD — get the z-score, percentile, and how rare the observation is (tail probability). Or go the other way: enter a z-score, get the matching raw value. Standard normal math, NIST-style.

Instant result
Private — nothing saved
Works on any device
AI insight included

Reviewed by CalcBold Editorial · Sources: NIST SEMATECH §1.3.5.6 (standard normal distribution) + standard statistical methodologyLast verified May 15, 2026Methodology

Embed builderDrop the Z-Score on your site →Free widget · 3 sizes · custom theme · auto-resizes · no signupGet embed code

Frequently Asked Questions

The most common questions we get about this calculator — each answer is kept under 60 words so you can scan.

What is a z-score?
A z-score (also called a standard score) measures how many standard deviations an observation is from the mean. z = (x − μ) / σ. Positive z = above mean; negative z = below; zero = at the mean. A z of 2 means the observation is 2 standard deviations above the mean — relatively unusual in a normal distribution (~2.3% of values).
What's a 'good' or 'bad' z-score?
Context-dependent. For test scores, percentile rank, or measurements where higher is better, positive z is good. For risk metrics, error rates, or where lower is better, negative z is good. As a general rule: |z| < 1 is typical (~68% of values), |z| < 2 is mildly unusual, |z| < 3 is rare, |z| ≥ 3 is very rare (<0.3% of normal data).
How does the percentile relate to z?
Percentile = the percentage of values that fall AT OR BELOW the observation. For a z-score of 1.0, the percentile is ~84.13% — 84% of normally-distributed values fall below that observation. The calculator uses the standard normal CDF (Φ(z)) to convert z to percentile.
What's the 'rarer than X% of values' line?
It tells you what fraction of observations are MORE extreme than yours in the tail direction. For a z-score of +2.5: 99.38% of values are below you; 0.62% are above. We report it as 'rarer than 0.62% of values' — meaning only 0.62% of the distribution is in the more-extreme tail. Useful for outlier interpretation.
Does z-score only work for normal distributions?
The CALCULATION (x − μ) / σ works for any distribution — it just measures how-many-SDs-from-mean. But the PERCENTILE interpretation (z = 1.0 → 84th percentile) is only correct for normal distributions. For skewed or heavy-tailed data, z = 1.0 might be at the 75th percentile or the 90th — the calculator's percentile assumes normality.
When is the value → z direction useful?
When you want to standardize observations across different scales or distributions. Convert SAT (μ=1050, σ=200) and ACT (μ=21, σ=5) to z-scores to compare student performance head-to-head. Convert blood pressure readings to z-scores against age-specific norms. Convert investment returns to z-scores against historical mean and SD. Standardization makes apples-to-apples comparison possible.
When is the z → value direction useful?
When you know a percentile and want the matching observed value. Examples: 'What height is the 95th percentile of US men?' → use μ=70 inches, σ=3, find x for z=1.645. 'What test score corresponds to the top 10%?' → use distribution parameters, find x for z=1.282. The reverse calculation is essential for setting thresholds.
What are critical z-values for common significance levels?
Two-tailed (most common): α=0.05 → ±1.960; α=0.01 → ±2.576; α=0.001 → ±3.291. One-tailed: α=0.05 → ±1.645; α=0.01 → ±2.326. Use these as decision thresholds in hypothesis testing — if |computed z| > critical z, reject the null hypothesis.
What's the empirical rule (68-95-99.7)?
For normally-distributed data: ~68.27% of values fall within ±1σ of the mean. ~95.45% within ±2σ. ~99.73% within ±3σ. Used as a quick mental check on observations: anything beyond ±3σ is rare enough that it suggests either an outlier or a non-normal distribution. The rule is approximate; the calculator's percentile is exact.
Can I have a z-score larger than 3?
Yes — z-scores have no upper or lower limit. A z of 4 is very rare in normal data (1 in 31,000); a z of 5 is essentially impossible (1 in 3.5 million) UNLESS the data is non-normal. Large z-scores almost always signal: (1) a data-entry error, (2) a non-normal distribution, or (3) a genuine outlier worth investigating.
How is the percentile calculated for negative z?
Same formula: percentile = Φ(z) × 100. For z = -1.5, percentile ≈ 6.68% — meaning the observation is below ~93% of values. The 'rarer than' line then reports it as 'rarer than 6.68% of values' — the smaller tail. Negative z is treated symmetrically with positive z; the underlying CDF is monotonic.
What if my σ is zero?
z-score is undefined when σ = 0 (all values identical; no spread). The calculator returns an error rather than dividing by zero. If you genuinely have a 'distribution' with σ = 0, there's nothing to standardize — every observation equals the mean by definition.

Z-Score Calculator

You might also need

Frequently Asked Questions