P(Z ≤ 1.96) = 0.9750. A z score table hands you that number in about five seconds — no integral, no software, just a row and a column. It’s a fixed grid of cumulative probabilities for the standard normal curve, and you’ll find it under several names: the standard normal z-table, the normal distribution table, the table of z-scores, or occasionally “z table normal” in a search box. All of them are the same grid.
The skill isn’t finding the table — it’s reading the right cell. Almost every mistake people make with a z score table comes down to one of three things: mixing up which digit belongs in the row versus the column, forgetting the table gives area below a value rather than at it, or misplacing the sign on a negative z. This guide works through the lookup mechanics in detail, including where practitioners get burned even after they’ve memorized the row-and-column rule, then hands you an interactive calculator to check every answer.
What Is the Z Score Table?
The z score table is a lookup grid for the standard normal distribution — a normal distribution with mean 0 and standard deviation 1, written N(0, 1). Every cell holds a cumulative left-tail probability: the probability that a standard normal random variable Z falls at or below the row-and-column z-value.
Table value = P(Z ≤ z) where Z ~ N(0, 1)
For z = 1.96, the table returns 0.9750 — 97.50 % of the standard normal distribution sits at or below that point. For z = −1.96 it returns 0.0250: only 2.50 % falls below there. The table was built before calculators existed, and it has stayed in every intro-stats course since, because working a lookup by hand teaches you the shape of the curve in a way that pressing a button on software does not. It also lets you sanity-check software output in seconds — if a script hands you a p-value of 0.98 for z = 1.96, you already know that’s wrong.
According to the NIST/SEMATECH e-Handbook of Statistical Methods, 1.3.6.1 — Normal Distribution, the standard normal distribution is the reference distribution behind z-tests, confidence intervals, normal approximations to the binomial, and process-capability analysis — which is why this one grid keeps reappearing across so much applied statistics.
The Standard Normal Distribution: A Quick Recap
Any normally distributed measurement X with population mean μ and population standard deviation σ converts to a z-score with:
z = (x − μ) / σ
The z-score counts how many standard deviations x sits above (positive) or below (negative) the mean. A z-score of +2.0 is two standard deviations above average; −1.5 is one and a half below.
Once you have z, the table — which always describes N(0, 1) — converts it into a probability. That’s why one table handles every normal distribution regardless of its original mean or spread: the standardization step absorbs both.
The curve is symmetric about 0, and its total area equals 1.00. The area beyond z = ±4 is smaller than 0.01 %, so most printed tables stop at z = ±3.49, covering more than 99.9 % of realistic outcomes.
How to Read the Table: Rows, Columns, Cells
Table Layout
- Rows carry the z-score to the tenths place — 1.9, 2.0, −0.5. Each row label is the whole number plus the first decimal digit.
- Columns carry the second decimal digit — 0.00, 0.01, … 0.09. The column header adds this hundredths amount onto the row’s value.
- Cells hold the cumulative probability P(Z ≤ z).
| Desired z | Row | Column |
|---|---|---|
| 1.96 | 1.9 | 0.06 |
| −1.25 | −1.2 | 0.05 |
| 0.83 | 0.8 | 0.03 |
| 2.14 | 2.1 | 0.04 |
Step-by-Step Lookup
To find P(Z ≤ 1.96):
- Locate the row labeled 1.9.
- Move across to the column labeled 0.06.
- Read the cell: 0.9750.
That single number says 97.50 % of the standard normal distribution falls at or below z = 1.96.
To find P(Z ≤ −0.83):
- Locate the row labeled −0.8.
- Move across to the column labeled 0.03.
- Read the cell: 0.2033.
About 20.33 % of the distribution lies below z = −0.83.
Reading the Column Headers Correctly
Each column header is the hundredths digit, not a step size. Column 0.06 means “the second decimal place of z is 6” — combined with row 1.9, that’s z = 1.96. Some older printed tables label columns 0 through 9 instead of 0.00 through 0.09; the logic is identical, just check the header before you read anything off.
A quick self-check: row label plus column label should equal your target z. Row 1.9 + column 0.06 = 1.96. If the sum doesn’t match what you’re looking for, you’re in the wrong row or column — this catches the single most common error before it costs you an answer.
Worked Example 1: Left-Tail Probability
Problem: Adult heights in a population are normally distributed with mean μ = 70 inches and standard deviation σ = 3 inches. What fraction of the population is shorter than 75.88 inches?
Step 1 — Convert to a z-score
z = (75.88 − 70) / 3 = 5.88 / 3 = 1.96
Step 2 — Look up z = 1.96
Row 1.9, Column 0.06 → P = 0.9750
Interpretation: 97.50 % of people in this population are shorter than 75.88 inches; only 2.50 % are taller. This particular z-value — 1.96 — shows up constantly because it’s also the critical value for a 95 % confidence interval: exactly 95 % of a standard normal distribution lies between −1.96 and +1.96.
Worked Example 2: A Quality-Control Reading
Problem: A machine shop cuts steel shafts meant to measure 25.00 mm in diameter. Historical data shows the cutting process is stable and normally distributed, with mean μ = 25.00 mm and standard deviation σ = 0.08 mm — typical CNC repeatability for this kind of part. An inspector pulls a shaft off the line and measures 24.90 mm. Should the floor supervisor be worried?
Step 1 — Convert to a z-score
z = (24.90 − 25.00) / 0.08 = −0.10 / 0.08 = −1.25
Step 2 — Look up z = −1.25
Row −1.2, Column 0.05 → P = 0.1056
Interpretation: about 10.56 % of shafts from this process run this small or smaller — roughly the 11th percentile. On its own, that’s an unremarkable reading from a process behaving normally, not evidence anything has drifted.
Now suppose the part spec has a hard lower limit of 24.84 mm — anything smaller gets scrapped. What fraction of the shop’s output falls below that line?
z = (24.84 − 25.00) / 0.08 = −0.16 / 0.08 = −2.00
Row −2.0, Column 0.00 → P = 0.0228
About 2.28 % of shafts from this process will fall outside the lower spec limit by chance alone, even with nothing wrong on the machine — a scrap rate worth flagging to engineering before it turns into a customer complaint. That’s the real value of the lookup: it turns a single measurement into a statement about the whole production run.
Try the Calculator
Enter any z-score below. The calculator returns the cumulative left-tail probability P(Z ≤ z), the right-tail probability P(Z > z), and the area between 0 and z — the same numbers a full printed table would give you, without the row-and-column hunting.
For a larger display with full options, open the z-table calculator. To explore other statistics tools, visit the calculators hub.
Right-Tail and Two-Tail Probabilities
The table gives left-tail probabilities, P(Z ≤ z), by default. Most real problems ask for something else — the area above a value, or the area between two values — so you need to convert.
Right-Tail: “Greater Than”
The area to the right of z is the complement of the left-tail area, since the total area under the curve is 1:
P(Z > z) = 1 − P(Z ≤ z)
Example: P(Z > 1.96) = 1 − 0.9750 = 0.0250. That 2.50 % right-tail area is the rejection threshold for a one-tailed hypothesis test at α = 0.05 — observing z > 1.96 leads to rejecting the null.
Two-Tail: “Between −z and +z”
Using the curve’s symmetry:
P(−z ≤ Z ≤ z) = 2 × P(Z ≤ z) − 1
Example with z = 1.96: P(−1.96 ≤ Z ≤ 1.96) = 2 × 0.9750 − 1 = 0.9500. Exactly 95.00 % of the distribution lies within ±1.96 standard deviations of the mean — the backbone of the two-sided 95 % confidence interval.
Between Two Different Z-Scores
For two different values z₁ < z₂:
P(z₁ ≤ Z ≤ z₂) = P(Z ≤ z₂) − P(Z ≤ z₁)
Example: P(0.50 ≤ Z ≤ 1.50) = 0.9332 − 0.6915 = 0.2417. About 24.17 % of the distribution falls between z = 0.50 and z = 1.50 — subtracting two lookups is the whole method.
Negative Z-Scores and the Symmetry Property
Negative rows trip up more learners than any other part of the table, even though the lookup process is identical — you just use the negative row in the left column. If your table only lists positive z-values, symmetry gives you a shortcut:
P(Z ≤ −z) = 1 − P(Z ≤ z)
Example: if your table shows P(Z ≤ 1.64) = 0.9495, then P(Z ≤ −1.64) = 1 − 0.9495 = 0.0505.
This holds because the curve is perfectly symmetric about zero: the left-tail area below −z equals the right-tail area above +z, which equals one minus the left-tail area below +z. It’s also why a symmetric two-tailed test only ever requires looking up the positive critical value — the negative one comes for free.
Two Table Formats You’ll Run Into — Don’t Mix Them Up
Not every printed table shows the same number for the same z, and confusing the two formats is the second most common table-reading error after swapping rows and columns.
| Format | What each cell holds | Range of values | Where you’ll see it |
|---|---|---|---|
| Cumulative (left-tail) | P(Z ≤ z) | 0 to 1 | Most modern textbooks; used throughout this page and the calculator above |
| 0-to-z (body) | P(0 ≤ Z ≤ z) | 0 to ~0.5 | Older or European textbooks; some engineering references |
The two are easy to convert between for a positive z:
Left-tail value = 0.5000 + body value
Body value = Left-tail value − 0.5000
Example: if a 0-to-z table gives you 0.4750 for z = 1.96, the cumulative equivalent is 0.5000 + 0.4750 = 0.9750 — matching the value used in every example above. Skipping this conversion is why two people can look up the same z-score and land 0.4750 apart on the answer while both being technically correct for their own table. Before trusting any printed grid, check its heading or footnote for which format it uses.
When the Z-Table Can Mislead You
The table itself is never wrong, but three assumptions behind it fail often enough in practice that they’re worth knowing before you lean on a lookup for a real decision.
You need a known σ, and you usually don’t have one. The formula z = (x − μ) / σ assumes σ is a true population parameter. In most real work you only have a sample standard deviation, s, estimated from limited data. Plugging an estimated s into the z formula and reading the z-table understates your uncertainty — the t-distribution, not the normal, is the statistically correct model whenever σ is estimated from a small sample. The rule of thumb is n < 30, though the t-distribution converges toward the normal as n grows, which is why software defaults to it even for larger samples.
The normality assumption itself can be wrong. No amount of table precision fixes a distribution that isn’t actually normal. Skewed data — hold times, income, reaction times — will give you a confidently wrong percentile if you standardize and look it up without checking a histogram or a Q-Q plot first.
Printed tables cap precision at two decimal places of z. Rounding z = 1.963 to 1.96 before lookup costs at most 0.0005 in probability — fine for coursework, not always fine for a regulatory report or a six-sigma capability study, where the exact z is what a calculator or statistical software should compute directly.
| Tool | Use it when | What it needs | Precision |
|---|---|---|---|
| Z-table | σ is genuinely known, or the sample is large enough to treat it as fixed | μ, σ | 4 decimals; z rounded to 0.01 |
| t-table | σ is estimated from a small sample | Degrees of freedom (n − 1) | Coarser; a handful of α levels per row |
| Calculator / software | Any case above, when you need an exact value | Same inputs as the matching table | Full floating-point precision |
This distinction matters most where the z-table shows up alongside a confidence interval — the interval’s width depends on picking the right one of the two distributions.
Applying the Table to Any Normal Distribution
The table always describes N(0, 1), but standardizing first lets you apply it to any normal variable:
- Identify μ and σ for the original variable X.
- Compute
z = (x − μ) / σfor the value you care about. - Look up z in the table to get P(Z ≤ z), which equals P(X ≤ x) in the original distribution.
One table, three steps, every normal distribution covered — it’s why textbooks pair the standardization formula with this table rather than printing a separate table for every possible mean and standard deviation.
For more practice with the standardization workflow, the Penn State STAT 414 course, Lesson 3.2 — The Standard Normal Distribution works through additional examples using the same method.
Common Mistakes When Reading the Table
Mixing Up Rows and Columns
Reading the column as the tenths digit instead of the hundredths digit is the single most common error. Row 1.9, column 0.06 is z = 1.96; row 1.6, column 0.09 is a completely different value, z = 1.69. Check yourself: row label plus column label should equal the z you’re after.
Forgetting the Table Gives Left-Tail Values
If a problem asks for the probability above a value, you have to subtract: P(Z > z) = 1 − P(Z ≤ z). Skipping that step turns a correct upper-tail answer of 0.0250 into an incorrect 0.9750.
Applying the Table to Non-Normal Data
The table is only valid when the underlying variable is normal (or close enough to approximate). Standardizing a strongly skewed distribution or a discrete count variable and looking up the result gives you a number that looks precise and means nothing.
Rounding the Z-Score Too Early
Round z = 1.963 to two decimal places (1.96), not one (2.0). Rounding to one decimal can cost you up to 0.0099 in probability — small, but avoidable.
Treating a Table Value as the Probability of One Exact Point
In a continuous distribution, P(Z = any exact value) is always zero. The table reports probabilities over intervals: P(Z ≤ 1.96) = 0.9750, but P(Z = 1.96) = 0.
Losing the Sign on a Negative Z-Score
Looking up z = 1.25 when you actually have z = −1.25 gives 0.8944 instead of the correct 0.1056 — the difference between the 89th percentile and the 11th.
Frequently Asked Questions
What is the z-score table used for?
It converts a standardized z-score into a cumulative probability — the fraction of a standard normal distribution at or below that value. That covers finding p-values for z-tests, locating critical values for confidence intervals, computing percentile ranks in a normally distributed population, and solving normal-probability problems without integrating by hand.
How do I find the probability for a z-score not listed in the table?
Round to two decimal places and use that entry — for z = 1.963, look up 1.96. The rounding error is at most 0.0005 in cumulative probability. For exact precision, use the calculator above, which computes the probability for any z-score directly rather than rounding to a grid.
What does a table value of 0.9750 mean?
That 97.50 % of the standard normal distribution falls at or below that z-value — equivalently, a random draw from N(0, 1) is less than or equal to z with probability 0.9750, leaving 2.50 % in the right tail.
What z-score corresponds to the 95th percentile?
Search the table for the value closest to 0.9500. You’ll find P(Z ≤ 1.64) = 0.9495 and P(Z ≤ 1.65) = 0.9505; interpolating gives the 95th-percentile z-score as approximately 1.645, commonly rounded to 1.64 or 1.65. The two-tailed 95 % critical value is different: ±1.96, since that’s the boundary for the middle 95 % rather than the bottom 95 %.
How do I use the table for a hypothesis test?
For a one-tailed z-test at significance level α, find z where P(Z ≤ z) = 1 − α. At α = 0.05, that’s 0.9500 → z ≈ 1.645; reject the null if your test statistic exceeds it. For a two-tailed test at α = 0.05, find z where P(Z ≤ z) = 0.9750 → z = 1.96; reject if |z| > 1.96.
Summary
Three ideas make every lookup in a z score table straightforward:
- Left-tail by default. Every cell gives P(Z ≤ z). Subtract from 1.00 for a right-tail probability; subtract two lookups for a range.
- Row for tenths, column for hundredths. Row 1.9 plus column 0.06 gives z = 1.96 — check that the sum matches your target z before trusting the cell.
- Standardize first. For any X ~ N(μ, σ), compute z = (x − μ) / σ before you open the table. One table then covers every normal distribution you’ll ever need.
Practice a few lookups — z = 1.00 (P = 0.8413), z = 1.64 (P ≈ 0.9495), z = −2.33 (P ≈ 0.0099) — until the row-and-column motion feels automatic, then check your answers against the calculator above.
To see how the normal distribution fits into the broader landscape of statistical distributions, read the probability distribution guide or browse every available tool on the calculators hub.