Z-Score Calculator

Formula

z = (x − μ) ÷ σ

z is the standard score; x is the individual value being evaluated; μ (mu) is the mean of the distribution; σ (sigma) is the standard deviation of the population.

Worked example

Value (x)

85

Mean (μ)

70

Standard deviation (σ)

10

• Z-score: 1.5
• Percentile (below x): 93.32%
• Above x: 6.68%

What Is a Z-Score and Who Uses It?

A z-score, also known as a standard score, measures the exact number of standard deviations a single data point lies above or below the mean of a dataset. By transforming raw measurements into a standardized scale where the mean is 0 and the standard deviation is 1, z-scores allow you to compare scores from entirely different metrics, scales, or testing environments.

Students, researchers, quality assurance engineers, medical clinicians, and financial analysts routinely use z-scores to:

• Compare standardized test results across different testing sessions or grading curves.
• Determine the relative rarity or typicality of biometric readings such as blood pressure or bone mineral density (DEXA scans).
• Identify statistical outliers in manufacturing and laboratory processes.
• Gauge relative market volatility and price deviations across financial assets.

---

Step-by-Step Worked Example

To see how the formula functions, consider an academic exam with the following established baseline:

• Value ($x$): 85
• Mean ($\mu$): 70
• Standard deviation ($\sigma$): 10

Step 1: Compute the Deviation from the Mean

Subtract the distribution mean from the individual observation: $$ x - \mu = 85 - 70 = 15 $$ A positive deviation indicates that the score is 15 points higher than the cohort average.

Step 2: Divide by the Standard Deviation

Scale this difference by the standard deviation to express the distance in standardized units: $$ z = \frac{15}{10} = 1.5 $$ The resulting standard score is 1.5.

Step 3: Determine the Cumulative Percentile

Using the standard normal cumulative distribution function (CDF):

• Percentile (below $x$): $\Phi(1.5) \approx 0.9332$ or 93.32%
• Probability above $x$: $1 - 0.9332 = 0.0668$ or 6.68%

An individual scoring 85 performed equal to or higher than 93.32% of all test-takers in this population, while only 6.68% achieved a higher score.

---

Comparing Different Scenarios

Because the z-score standardizes observations against the same mean (70) and standard deviation (10), shifting the raw input reveals dramatic changes in position along the normal curve:

| Input Level | Value ($x$) | Z-Score | Percentile (Below $x$) | Interpretation | | :--- | :--- | :--- | :--- | :--- | | Lower input | 68 | -0.2 | 42.07% | Slightly below average; well within normal bounds | | Base example | 85 | 1.5 | 93.32% | Above average; upper quartile performance | | Higher input | 102 | 3.2 | 99.93% | Extreme outlier; higher than 99.9% of population |

---

How to Read Z-Score Results and Established Benchmarks

Interpreting a z-score relies heavily on the Empirical Rule (or 68–95–99.7 rule) for standard normal distributions:

• Between -1.0 and +1.0 (68.27% of data): Represents average, typical outcomes. Scores in this band are common and expected.
• Between -2.0 and +2.0 (95.45% of data): Captures the vast majority of ordinary observations. A score of $\pm 1.96$ marks the conventional 95% threshold used in two-tailed hypothesis testing.
• Between -3.0 and +3.0 (99.73% of data): Encompasses almost all standard variability. Values falling between 2.0 and 3.0 are considered uncommon.
• Beyond $\pm 3.0$ (0.27% of data): Classified as extreme statistical outliers. Observations at this level warrant scrutiny for data entry errors, special experimental causes, or unique real-world conditions.

---

Practical Tips for Accurate Calculations

1. Maintain Consistent Units: Ensure the individual observation, mean, and standard deviation share identical units (e.g., all in pounds, inches, or raw points). Never mix scaled metrics.
2. Check the Sign: A negative z-score is normal and mathematically valid. It simply indicates that the observation is below the group average.
3. Distinguish Sample from Population: When working with sample standard deviations ($s$) rather than population parameters ($\sigma$) on small datasets ($n < 30$), transition to a t-distribution to prevent underestimating variance.

---

When Not to Rely on the Z-Score

While the mathematical calculation $z = (x - \mu) / \sigma$ can technically be applied to any numerical data, relying on it for percentiles and probability areas requires an approximately normal distribution. Avoid relying on standard normal conversions under the following conditions:

• Heavily Skewed Data: Distributions like personal income, real estate prices, or website pageviews are strongly right-skewed. Applying z-score percentiles to skewed datasets produces misleading probability estimates.
• Bimodal or Multimodal Curves: If the data features multiple peaks (such as combined male and female height distributions), standard deviation does not represent a single central tendency.
• Extreme Fat-Tailed Distributions: In financial market returns or catastrophe modeling, tail risks occur far more frequently than predicted by a Gaussian bell curve. In these contexts, standard z-score probabilities underestimate the true frequency of extreme events.

Comparison

Scenario	Input	Result
Lower input	Value (x): 68	Z-score: -0.2
Base example	Value (x): 85	Z-score: 1.5
Higher input	Value (x): 102	Z-score: 3.2

Common mistakes

• Subtracting the value from the mean instead of the mean from the value, which reverses the sign of the score.
• Using the standard error instead of the standard deviation when analyzing individual observations rather than sample means.
• Applying z-score percentile tables to non-normal, heavily skewed, or bimodal distributions.
• Confusing negative z-scores with negative probabilities or negative raw values.
• Dividing by variance instead of standard deviation.

Checklist

• Identify the raw value (x) you want to standardize.
• Verify whether you are using a population mean (μ) or sample mean (x̄).
• Confirm the standard deviation is strictly greater than zero.
• Ensure the underlying data distribution is approximately normal or bell-shaped.
• Express all three variables (x, μ, σ) in identical units of measurement.
• Determine whether you need the cumulative area below x, above x, or two-tailed.

Calculators Percentage Editorial Team

Calculator guides from the Calculators Percentage team. These introductory examples are educational, not professional advice.

Last reviewed: November 12, 2026

This calculator is designed for statistical analysis and educational purposes. It assumes an approximately normal distribution for percentile conversions and should not replace formal statistical modeling in clinical, legal, or high-stakes financial decisions.