2026-08-07 10 min

Z-Score Calculator

Convert raw scores to z-scores from mean and standard deviation, with left/right-tail p-values, percentiles, T-scores, and between-two-values probabilities.

Calculator

Single Score

Raw Score (x)

Mean (μ)

Std Dev (σ)

Between Two Values

Lower Value (x₁)

Upper Value (x₂)

IQ Reference (μ=100, σ=15)

IQ	z	Percentile
55	-3.00	0.13%
70	-2.00	2.28%
85	-1.00	15.87%
100	0.00	50.00%
115	1.00	84.13%
130	2.00	97.72%
145	3.00	99.87%
160	4.00	100.00%

Results update automatically

Results update live as you type

Status:z = 1.5093.3th percentileP between 0.6827

Z-Score Summary

Z-Score1.5000

Left-Tail P0.9332

Right-Tail P0.0668

Percentile93.32%

P(60 < X < 80)0.6827

Score Breakdown

T-Score (50 + 10z)65.00

z₁ (lower)-1.0000

z₂ (upper)1.0000

SDs from mean1.50σ above

Results

Normal

Z-Score

1.5000

(x − μ) / σ

Left-Tail P (Cumulative)

0.9332

P(X ≤ x): area left of the score

Right-Tail P

0.0668

P(X > x) = 1 − left-tail

Percentile Rank

93.32%

Share of population scoring lower

P(60 < X < 80)

0.6827

Bell area between the two values

T-Score

65.00

50 + 10z; mean-50, SD-10 scale

Z-scores divide by the entered standard deviation, which must be positive. Tail and between probabilities difference the error-function normal CDF and assume approximately normal data.

Visualizations

Cumulative Probability at Reference z

0z=-2

0.2z=-1

0.5z=0

0.8z=1

1z=2

Tail Split at x

Left tail93%

Right tail7%

Percentile Rank

Percentile93.3%

Between-Values vs Total Area

Between68.3%

Formula

The z transformation recenters any normal distribution to mean 0, spread 1, so a single standard CDF answers every tail, percentile, and between-values question. T-scores relabel the same information onto the familiar 50±10 scale.

z = (x − μ) / σ

Alternative Form

Φ(z) = ½[1+erf(z/√2)] · P(a<X<b) = Φ(z₂)−Φ(z₁) · T = 50+10z · percentile = 100×Φ(z)

Variables

Symbol	Name	Unit	Description
x	Raw score	data	Observed value to standardize
μ	Population mean	data	Distribution center
σ	Standard deviation	data	Spread; must be positive
z	Z-score	σ	Signed SDs from the mean
Φ	Standard normal CDF	probability	Left-tail area at z
T	T-score	T	50 + 10z clinical scale
Pₐ_b	Between probability	probability	Area between two raw values

Explanation

Standardization separates shape from location: every normal question becomes one lookup on N(0,1). Percentiles, tails, and between-areas are all just CDF readings at the right z.

Derivation

Linearity of expectation and variance under (x−μ)/σ maps N(μ,σ²) to N(0,1). The Gaussian integral has no elementary antiderivative, so Φ is evaluated through the erf series approximation; T is the affine map sending z=0→50 with slope 10.

Step-by-Step Calculation

Follow each step to understand how the result is calculated from your inputs.

Step 1

Validate the raw score, mean, and a positive standard deviation.

require σ > 0

x:85

μ:70

σ:10

Intermediate: finite numbers with positive spread

Result: Inputs valid

Step 2

Standardize the raw score into z units.

z = (x − μ) / σ

x:85

μ:70

σ:10

Intermediate: (85 − 70) / 10

Result: z = 1.5000

Step 3

Evaluate the standard normal CDF at z for the left-tail probability.

Φ(z) = ½[1 + erf(z/√2)]

z:1.5000

Intermediate: error-function approximation

Result: P(X ≤ x) = 0.9332

Step 4

Complement for the right tail and scale by 100 for the percentile.

right = 1 − Φ(z); percentile = 100×Φ(z)

z:1.5000

Intermediate: 1 − 0.9332

Result: 93.32th percentile

Step 5

Standardize both between-values bounds.

z₁ = (x₁−μ)/σ; z₂ = (x₂−μ)/σ

x₁:60

x₂:80

Intermediate: z₁ = -1.0000, z₂ = 1.0000

Result: Bounds standardized

Step 6

Difference the CDFs for the between-values area.

P = |Φ(z₂) − Φ(z₁)|

x₁:60

x₂:80

Intermediate: absolute difference (order-free)

Result: P = 0.6827

Final Answer

1.5000

Worked Examples

See how the calculator works with different scenarios.

Scenario

Score x=85 with class mean 70 and SD 10.

Inputs

x:85

μ:70

σ:10

Calculation

z = (85−70)/10 = 1.50 Φ(1.5) ≈ 0.9332

Result

z = 1.50 · 93.32nd percentile

Interpretation

One-and-a-half SDs above average beats roughly 93 of 100 classmates.

Understanding Your Result

Your Result: Z-Score: 1.5000

Typical range: |z| < 2 covers the central ~95%; |z| > 2 deserves attention; |z| > 3 is outlier territory.

What This Means

The z-score is a universal ruler: +1.5 means the same relative standing on any normal test, measurement, or process.

Why It Matters

Comparing raw scores across different scales is meaningless; comparing z-scores (or percentiles) is fair — the basis of admissions, QC limits, and clinical cutoffs.

Professional Interpretation

Quote z with the assumed μ and σ, and confirm normality before converting z to probabilities.

Recommendations

•Always report μ and σ with z
•Check normality before probability claims
•Use |z| > 3 as an outlier tripwire
•Prefer two-tailed framing unless directed
•Round z to 2 decimals, p to 4
•Convert to percentiles for lay audiences

Educational Content

Introduction

Z-scores translate any normal measurement into standard-deviation units, unlocking tail probabilities, percentiles, and fair cross-scale comparisons from one formula.

Concept Explanation

Subtract the center, divide by the spread: the result says how exceptional a value is in a language every normal distribution shares.

Applications

▸Standardized testing (SAT, GRE, IQ)

▸Quality-control specification limits

▸Clinical reference ranges and growth charts

▸Finance: Sharpe-style risk standardization

▸Sports analytics across eras

▸Hypothesis-test statistics

Benefits

•Fair comparison across different scales
•Instant percentiles and tail areas
•Outlier detection with principled cutoffs
•T-score output for clinical contexts
•Between-values areas without tables

Limitations

•Probability readings assume normality
•Sensitive to misestimated μ and σ
•Single-value view hides distribution shape
•erf approximation error ~1e−7

Best Practices

•Verify approximate normality first
•Use robust μ/σ with contaminated data
•State one- vs two-tailed framing
•Keep full precision until reporting
•Pair z with the raw score for context

Common Misconceptions

Myth: Z-scores only work on bell-shaped data.

Truth: The arithmetic always works; only the probability interpretation needs normality.

Myth: A high percentile means a high z.

Truth: Percentiles saturate: z = 2 → 97.7th, but z = 3 → 99.87th — tiny percentile moves, huge rarity moves.

Myth: Negative z is bad.

Truth: Sign only encodes direction from the mean — below-average cholesterol z is good news.

Myth: T-scores are a different test.

Truth: T = 50+10z is the same information relabeled; T = 60 is exactly z = 1.

Practical Tips

•Memorize z = 1.96 ↔ 95% two-tailed
•Sketch the curve before computing
•Use between-mode for “within” questions
•Check σ > 0 when results blank
•Convert z to percentiles for reports
•Compare IQ-style scores via z, not raw
•Keep 4 decimals on probabilities

Common Mistakes

Dividing by variance instead of SD

z uses σ (square root of variance) — dividing by variance understates z.

Assuming normality blindly

Histogram the data; skew breaks tail-probability readings.

One-tailed vs two-tailed mix-ups

Double one-tailed p for two-tailed tests; state the choice.

Reading percentile as percent correct

93rd percentile means beating 93% of people, not scoring 93%.

Using sample SD as σ silently

Label estimates as estimates; tiny samples give noisy z.

Assumptions

This calculator makes the following assumptions to provide estimates:

Known Parameters: μ and σ are treated as the true population values.

Normality for P: Tail and between probabilities assume Gaussian data.

Positive Spread: σ must exceed zero.

Single Observation: z describes one value, not a sample mean (no √n).

Numerical CDF: erf approximation backs every probability.

Limitations

Be aware of these limitations when using this calculator:

Normal-Only Probabilities: t, chi-square, and skew distributions need other tools.

No Sample-Mean Mode: Standard error (σ/√n) framing is out of scope.

Parameter Sensitivity: Wrong μ/σ shifts every downstream number.

No Hypothesis Tests: Critical values and p-decisions are not automated.

Frequently Asked Questions

Answers to common questions about this calculator.

What is a z-score?

A standard score measuring how many standard deviations a value sits from the mean: z = (x − μ) / σ. Positive means above the mean; negative means below.

How do you calculate a z-score?

Subtract the mean, divide by the standard deviation. For x = 85, μ = 70, σ = 10: z = (85−70)/10 = 1.50.

What does a z-score of 1.5 mean?

The score is 1.5 SDs above the mean — cumulative probability ≈ 0.9332, higher than about 93.3% of the population.

What is the difference between z-score and standard deviation?

Standard deviation describes a dataset’s spread; a z-score locates one value within that spread, using the SD as its unit.

What is the empirical rule (68-95-99.7)?

For normal data ≈68% falls within ±1σ, 95% within ±2σ, 99.7% within ±3σ of the mean — z-scores of 1, 2, and 3 mark those gates.

How do you interpret a negative z-score?

The value is below the mean. z = −0.5 leaves a left-tail area of ≈0.3085: about 30.9% of values fall lower.

How do you find the percentile from a z-score?

Take the left-tail cumulative probability × 100. Cumulative 0.9332 → 93.32nd percentile.

What does a z-score of 0 mean?

The raw score equals the mean exactly — 50th percentile, half the population below and half above.

How do you calculate the probability between two z-scores?

Difference the CDFs: P(z₁ < Z < z₂) = Φ(z₂) − Φ(z₁) — the bell area between the two markers.

What is a T-score and how does it relate to z-scores?

T = 10z + 50 recenters scores to mean 50, SD 10 (common in clinical testing). T = 60 equals z = 1.0.

How do z-scores help detect outliers?

|z| > 3 flags only ~0.3% of normal data, so such points are investigated as potential outliers; |z| > 4 is near-certain.

Can z-scores be used for non-normal distributions?

They can be computed, but probability readings assume normality. Otherwise Chebyshev’s inequality bounds coverage: ≥75% within |z| < 2, ≥89% within |z| < 3.