Z-Score to P-Value: How to Read the Normal Distribution

A p-value is simply the area under the normal curve beyond your z-score. Convert a z to a p by taking a tail: right-tailed is P(Z > z), left-tailed is P(Z < z), and two-tailed is 2 × P(Z > |z|). The two numbers everyone remembers: z = 1.96 → two-tailed p = 0.05, and z = 1.645 → one-tailed p = 0.05.

What a z-score and a p-value each mean

A z-score answers “how unusual is this value?” in units of standard deviations:

z = (x − μ) / σ

A z of +1 sits one standard deviation above the mean; a z of −2.5 sits two-and-a-half below. The p-value turns that position into a probability: the chance of landing at least that far out in the tail if nothing but random variation were at work. Position in, probability out — that is the whole conversion, and the normal distribution calculator shows it with the tail shaded so you can see the area you are measuring.

The three tails

The one detail that trips people up is which area to take.

• Right-tailed — you only care about values larger than expected: p = P(Z > z).
• Left-tailed — you only care about values smaller: p = P(Z < z).
• Two-tailed — you care about any difference in either direction: p = 2 × P(Z > |z|).

Because the normal curve is symmetric, the two-tailed p-value is exactly double the one-tailed value. That is why one significance level, 0.05, maps to two different critical z-scores: 1.645 if you are testing in one direction, 1.96 if you are testing in both.

The values worth memorising

Confidence	α	Two-tailed z	One-tailed z
90%	0.10	1.645	1.282
95%	0.05	1.960	1.645
99%	0.01	2.576	2.326

Read it either way: a z of 1.96 gives a two-tailed p of 0.05, and a 95% confidence interval uses that same 1.96. The confidence interval calculator uses these identical critical values to build an interval around an estimate.

Why a table is not enough

The normal cumulative distribution function — the “area so far” under the curve — has no elementary closed form. Every exact tool computes it through the error function (erf), a high-accuracy series that pins the area to about seven decimal places. A printed z-table, by contrast, is rounded to two decimals of z and forces you to interpolate between cells; and a chatbot asked for P(Z > 2.13) will often quote a plausible-but-wrong figure. Computing it directly avoids both problems — enter any z into the p-value calculator and it returns the exact tail area (and it also handles t, χ² and F statistics, not just z).

Turning the p-value into a decision

A p-value on its own is just an area. It becomes a decision when you compare it to your significance level α — the threshold you set before looking at the data, conventionally 0.05.

• p ≤ α → the result is statistically significant; you reject the null hypothesis.
• p > α → not significant; you fail to reject it.

A small p-value means your statistic sits far out in the tail — the kind of value that would rarely occur by chance alone. It does not measure the size or importance of an effect, only how surprising the data would be if the null hypothesis were true.

Quick summary

To convert a z-score to a p-value, take the area under the normal curve beyond it: right-tailed P(Z > z), left-tailed P(Z < z), or two-tailed 2 × P(Z > |z|). Remember z = 1.96 → p = 0.05 (two-tailed) and z = 1.645 → p = 0.05 (one-tailed), compare the p-value to your α, and use a tool that computes the exact area via erf rather than a rounded table. Try it on the normal distribution calculator or the p-value calculator.

Sources: the standard normal distribution and its cumulative distribution function (the error function, erf); conventional significance testing as taught in introductory statistics. Educational information — not statistical consulting.