Skip to content
Statculator
P-Values & Critical Values

T Critical Value Calculator

Free t critical value calculator and t table. Get t* for any degrees of freedom (decimals too) and α, one- or two-tailed, with steps and a shaded t curve.

Free, no sign-up Runs in your browserBy Hira Noor, BS MathematicsUpdated

Try

Any positive value, including decimals such as a Welch df of 9.37.

Usually 0.05 or 0.01.

Tail

Two-tailed for H₁: “not equal to” and two-sided confidence intervals; one tail for directional tests.

Results

Critical value t* (two-tailed)
±2.086
df = 20, α = 0.05
z* at the same α, for comparison
±1.96
Degrees of freedom
20
Significance level, α
0.05
Area in each tail
0.025
Cumulative area, P(T ≤ t*)
0.975
How much larger t* is than z*
6.43%
Decision rule
Reject H₀ if |t| ≥ 2.086

Interpretation: With 20 degrees of freedom, reject H₀ in a two-tailed t-test at α = 0.05 when |t| ≥ 2.086. That cut-off is 6.43% farther from 0 than the normal value ±1.96, the price of estimating the standard deviation from the sample.

t distribution curve with the shaded probability region, markers at -2.086, t* = 2.086-2.086t* = 2.086-4-3-2-101234t distribution, df = 20
Shaded: rejection region with total area α = 0.05 (0.025 in each tail). Dashed lines: critical values.

t critical values for df = 20

Confidence (two-sided)α two-tailedα one-tailedt*z*
80%0.20.11.32531.2816
90%0.10.051.72471.6449
95%0.050.0252.0861.96
98%0.020.012.5282.3263
99%0.010.0052.84532.5758
99.8%0.0020.0013.55183.0902
99.9%0.0010.00053.84953.2905

Read across a row: the same t* serves a two-tailed test at α, a one-tailed test at α/2 and a two-sided interval at the stated confidence. Use −t* for a left-tailed test. The z* column shows the normal value the t* converges to as df grows.

Step-by-step solution

  1. 1

    Decide where α goes

    A two-tailed test splits α equally between the two tails.

    α/2 = 0.05 / 2 = 0.025 in each tail
  2. 2

    Find the cumulative area to the left of t*

    P(T ≤ t*) = 1 − α/2 = 0.975
  3. 3

    Invert the t CDF with df = 20

    There is no closed form, so the inverse is found numerically: the value whose cumulative t probability equals the target.

    t* = T⁻¹(0.975; df = 20) = 2.086
    By symmetry the lower value is −2.086, so t* = ±2.086
  4. 4

    Compare with the normal critical value

    The t distribution has heavier tails, so t* is always farther from 0 than z* at the same α.

    z* = ±1.96, t* / z* = 1.0643
  5. 5

    Use the critical value

    Reject H₀ if |t| ≥ 2.086
    95% confidence interval: estimate ± 2.086 × SE

What is a T Critical Value Calculator?

A t critical value calculator finds t*, the critical t value that leaves area α in the tail or tails of a Student t distribution with a given number of degrees of freedom. For a two-tailed test or a two-sided interval, t* = T⁻¹(1 − α/2; df), so 95% confidence with 20 df gives ±2.086. It works for any positive df, including the decimal df from Welch’s test.

What the T Critical Value Calculator does

Enter the degrees of freedom, the significance level and the tail. The calculator returns t* with the rejection region shaded on the t curve, the matching normal value z* for comparison, and a full row of a t-table for your df: t* at 80%, 90%, 95%, 98%, 99%, 99.8% and 99.9% confidence. It replaces a printed t critical value table, because it works for any df, including decimal df.

The same number goes by several names: the critical t value, the t-score critical value, the t statistic critical value, t star, t-crit or t_α/2. Whatever the name, it is the cut-off your t statistic is compared with, and the multiplier in a confidence interval.

You need t* for every confidence interval or test about a mean when σ is unknown: one-mean intervals, one-sample, two-sample, Welch and paired t-tests, and intervals for regression coefficients. Unlike a printed t-table, it does not skip from df = 30 to 40 to 60 and it handles the fractional df that Welch’s method produces.

How to use the T Critical Value Calculator

  1. Enter the Degrees of freedom (df), for example n − 1 for a one-sample interval. The Degrees of Freedom Calculator covers other designs.
  2. Enter the Significance level (α). For a confidence level C, use α = 1 − C (95% → 0.05).
  3. Choose the Tail: Two-tailed for “not equal to” tests and two-sided intervals, or Right / Left for a directional test.
  4. Read t* and compare it with z*. The table gives t* for other confidence levels at the same df.

How to find the t critical value

To calculate the t critical value by hand you need three things: the degrees of freedom, the significance level α and the tail of the test. Then read t* from a t table or compute the t quantile.

  1. Work out df. One sample or paired data: n − 1. Pooled two-sample test: n₁ + n₂ − 2. Welch’s test: the Welch–Satterthwaite value, usually a decimal.
  2. Choose α. 0.05 is the usual default; for a confidence level C, α = 1 − C.
  3. Pick the tail. Two-tailed puts α/2 in each tail; one-tailed puts all of α in one tail.
  4. Look up or compute the critical value. In the t critical value table on this page, go to the row for your df and the column for your α. Or compute T⁻¹(1 − α/2; df) for two-tailed and T⁻¹(1 − α; df) for one-tailed. Example: df = 20, α = 0.05 gives ±2.086 two-tailed and 1.725 right-tailed.

One-tailed or two-tailed: which t critical value do you need?

Choose the tail from the alternative hypothesis, before you look at the data:

  • Two-tailed (H₁: μ ≠ μ₀): use ±t*, with α/2 in each tail. Also use the two-tailed value for any two-sided confidence interval.
  • Right-tailed (H₁: μ > μ₀): use +t*, with all of α in the upper tail. Reject when t ≥ t*.
  • Left-tailed (H₁: μ < μ₀): use −t*, with all of α in the lower tail. Reject when t ≤ −t*.
  • A one-tailed t* is always smaller than the two-tailed t* at the same α (1.725 versus 2.086 at df = 20), so pick the tail on the research question, not to get a smaller cut-off.
  • Not a t problem? Use the z critical value when σ is known, the chi-square critical value for counts and variances, the F critical value for ANOVA, or the general Critical Value Calculator to switch between all four.

T critical value table

This table of t critical values (a t distribution table, or t test critical value chart) gives right-tail values: each entry leaves the column’s one-tailed α in the upper tail of t(df). For a two-tailed test, use the second number in each column heading, which is the two-tailed α. Use −t* for a left-tailed test. For any df not listed, including decimals, use the calculator above. The full t-table has more rows and a free PDF.

Critical values of t, by degrees of freedom and α
dfOne-tailed 0.10 · two-tailed 0.200.05 · 0.100.025 · 0.050.01 · 0.020.005 · 0.010.001 · 0.002
13.0786.31412.70631.82163.657318.309
21.8862.9204.3036.9659.92522.327
31.6382.3533.1824.5415.84110.215
41.5332.1322.7763.7474.6047.173
51.4762.0152.5713.3654.0325.893
61.4401.9432.4473.1433.7075.208
71.4151.8952.3652.9983.4994.785
81.3971.8602.3062.8963.3554.501
91.3831.8332.2622.8213.2504.297
101.3721.8122.2282.7643.1694.144
111.3631.7962.2012.7183.1064.025
121.3561.7822.1792.6813.0553.930
131.3501.7712.1602.6503.0123.852
141.3451.7612.1452.6242.9773.787
151.3411.7532.1312.6022.9473.733
161.3371.7462.1202.5832.9213.686
171.3331.7402.1102.5672.8983.646
181.3301.7342.1012.5522.8783.610
191.3281.7292.0932.5392.8613.579
201.3251.7252.0862.5282.8453.552
211.3231.7212.0802.5182.8313.527
221.3211.7172.0742.5082.8193.505
231.3191.7142.0692.5002.8073.485
241.3181.7112.0642.4922.7973.467
251.3161.7082.0602.4852.7873.450
261.3151.7062.0562.4792.7793.435
271.3141.7032.0522.4732.7713.421
281.3131.7012.0482.4672.7633.408
291.3111.6992.0452.4622.7563.396
301.3101.6972.0422.4572.7503.385
401.3031.6842.0212.4232.7043.307
601.2961.6712.0002.3902.6603.232
801.2921.6641.9902.3742.6393.195
1001.2901.6601.9842.3642.6263.174
1201.2891.6581.9802.3582.6173.160
∞ (z*)1.2821.6451.9602.3262.5763.090

T critical value formula

Each critical value is a quantile of the t distribution with ν degrees of freedom, found by inverting its CDF numerically. There is no closed-form formula for t*.

Two-tailed (two-sided interval)
t* = T⁻¹(1 − α/2; ν) → reject if |t| ≥ t*
Right-tailed
t* = T⁻¹(1 − α; ν) → reject if t ≥ t*
Left-tailed
t* = T⁻¹(α; ν) = −T⁻¹(1 − α; ν) → reject if t ≤ t*
Confidence interval for a mean
x̄ ± t* × s / √n, ν = n − 1
Formula symbols
SymbolMeaning
t*critical value (also written t_α/2, ν)
ν (df)degrees of freedom
αsignificance level
T⁻¹(p; ν)inverse CDF (quantile function) of the t distribution

As ν → ∞ the t distribution becomes the standard normal, so t* → z*. At ν = 120 the two-tailed 5% value is already 1.9799, within about 1% of 1.96.

Sources: NIST/SEMATECH e-Handbook, 1.3.6.7.2 Critical values of Student’s t distribution; NIST/SEMATECH e-Handbook, 1.3.6.6.4 t distribution; Microsoft Support, T.INV.2T function. Values on this page were checked against SciPy’s t.ppf.

Worked example: two-tailed t-test

Is mean body temperature 98.6 °F?

Problem. A sample of n = 25 body temperatures has mean 97.9 °F and standard deviation 1.5 °F. Test H₀: μ = 98.6 against H₁: μ ≠ 98.6 at α = 0.05.

  1. 01df = n − 1 = 24. The test is two-tailed, so α/2 = 0.025 goes in each tail and the cumulative area to the left of t* is 0.975.
  2. 02Critical value: t* = T⁻¹(0.975; 24) = ±2.0639. In the t table, read row 24, column “0.025 · 0.05”: 2.064.
  3. 03Test statistic: t = (97.9 − 98.6) / (1.5 / √25) = −0.7 / 0.3 = −2.333.
  4. 04|t| = 2.333 is larger than 2.0639, so t lands in the lower rejection region.

Answer. Reject H₀ at α = 0.05 (p ≈ 0.028). The 95% confidence interval, 97.9 ± 2.0639 × 0.3 = (97.28, 98.52), also excludes 98.6.

Worked example: one-tailed t-test

A sample of n = 16 has mean 52.1 and standard deviation 4.8. Test H₀: μ = 50 against H₁: μ > 50 at α = 0.05.

  1. df = n − 1 = 15. The test is right-tailed, so all 5% goes in the upper tail.
  2. Critical value: t* = T⁻¹(0.95; 15) = 1.7531.
  3. Test statistic: t = (52.1 − 50) / (4.8 / √16) = 2.1 / 1.2 = 1.75.
  4. Decision: 1.75 is just below 1.7531, so do not reject H₀ (p ≈ 0.0503). A two-tailed test would need |t| ≥ 2.1314, so it would not reject either.

Worked example: Welch’s fractional degrees of freedom

A Welch two-sample comparison gives df = 9.37. Find the two-tailed critical value at α = 0.05 and compare it with a printed table.

  1. Split α: α/2 = 0.025 in each tail, so the cumulative area to the left of t* is 0.975.
  2. Invert the t CDF with 9.37 df: t* = T⁻¹(0.975; 9.37) = 2.2486.
  3. A printed table has no row for 9.37 and would be read at df = 9, giving 2.2622, slightly more conservative.
  4. Reject H₀ if |t| ≥ 2.2486. That is about 14.7% farther from 0 than z* = 1.96.

t* versus z* as degrees of freedom grow

With few degrees of freedom the sample standard deviation is a shaky estimate, so t* is much larger than z* to keep the error rate at α. The gap closes quickly: by 30 df the two-tailed 5% value is only about 4% above 1.96.

dfTwo-tailed t* (α = 0.05)One-tailed t* (α = 0.05)Larger than z* by
112.70626.3138548%
24.30272.9200120%
33.18242.353462%
52.57062.015031%
102.22811.812513.7%
202.08601.72476.4%
302.04231.69734.2%
602.00031.67062.1%
1201.97991.65771.0%
∞ (z*)1.96001.64490%

T critical values in Excel, R, Python and on a TI-84

Most functions take a left-tail (cumulative) probability and df. Excel’s T.INV.2T is the exception: it takes α itself and returns the positive two-tailed value. The examples use df = 20 and α = 0.05, which return 2.086 two-tailed and 1.725 one-tailed.

ToolTwo-tailed t*One-tailed t* (right)
Excel / Google Sheets=T.INV.2T(0.05, 20)=T.INV(0.95, 20)
TI-84invT(0.975, 20)invT(0.95, 20)
Rqt(0.975, 20)qt(0.95, 20)
Python (SciPy)t.ppf(0.975, 20)t.ppf(0.95, 20)

Common use cases

  • Building a confidence interval for a mean with an estimated standard deviation.
  • Setting the rejection region for any t-test before running it.
  • Finding the multiplier for intervals on regression slopes and predictions, with n − k − 1 df.
  • Handling Welch and Satterthwaite fractional df that tables cannot.

Assumptions and limitations

  • t* gives correct error rates only when the underlying data (or differences) are roughly normal, or the sample is large enough for the central limit theorem to help.
  • The df must match the procedure: n − 1 for one sample, n₁ + n₂ − 2 for a pooled test, the Welch–Satterthwaite value for Welch’s test.
  • If σ is genuinely known, the normal critical value from the Z Critical Value Calculator is the right one, not t*.
  • The critical value only sets the cut-off. For the exact p-value of your t statistic, use the P-Value from t Calculator.

Frequently asked questions

What is a t critical value?

A t critical value (t*) is the cut-off on a Student t distribution that leaves a chosen area α in the tail or tails. In a t-test you reject H₀ when your t statistic is beyond t*; in a confidence interval t* is the multiplier on the standard error. It depends on the degrees of freedom, α and whether the test is one- or two-tailed.

How do you find the t critical value?

You need the degrees of freedom, the significance level α and the tail. For a two-tailed test take the t value that leaves α/2 in each tail, for example t* = 2.2281 for df = 10 at α = 0.05; for a one-tailed test put all of α in one tail, which gives 1.8125 for df = 10. This calculator returns t* for any df, so you do not need a printed t-table.

How do you calculate the t critical value without a table?

Compute the inverse t CDF: t* = T⁻¹(1 − α/2; df) for a two-tailed test or T⁻¹(1 − α; df) for a one-tailed test. There is no closed-form formula, so software does it numerically: qt(0.975, 20) in R, =T.INV.2T(0.05, 20) in Excel or invT(0.975, 20) on a TI-84 all return 2.086.

What is the t critical value for a 95% confidence interval?

It depends on df. For a two-sided 95% interval, t* is 12.706 at 1 df, 2.776 at 4 df, 2.228 at 10 df, 2.086 at 20 df, 2.042 at 30 df and approaches 1.96 for very large df. Enter your df above to get the exact value.

What is the one-tailed t critical value at α = 0.05?

It is the value with 5% of the area to its right: 1.812 at 10 df, 1.725 at 20 df and 1.697 at 30 df, approaching 1.645 for large df. Use the negative of that value for a left-tailed test.

How do I find the t critical value on a TI-84?

Press 2nd, VARS (DISTR) and choose invT(. Enter the cumulative area to the left and the df: invT(0.975, 20) gives 2.086 for a two-tailed test at α = 0.05, and invT(0.95, 20) gives 1.725 for a right-tailed test. Older TI-83 models have no invT, so use this calculator or a t table instead.

How do I find degrees of freedom for a t critical value?

For a one-sample or paired t procedure, df = n − 1. For a pooled two-sample test, df = n₁ + n₂ − 2. For Welch’s test, use the Welch–Satterthwaite formula, which usually gives a fractional value. The Degrees of Freedom Calculator computes all of these.

Can the degrees of freedom be a decimal?

Yes. The t distribution is defined for any positive df, and Welch’s test routinely produces values like 9.37. This calculator uses the exact df; a printed table forces you to round down, which gives a slightly larger, more conservative t*.

Why is t* bigger than z*?

Using the sample standard deviation instead of the true σ adds uncertainty, which the t distribution reflects with heavier tails. To keep the same α, the cut-off must move farther from 0. The difference shrinks as df grows and the estimate of σ improves.

Cite this page

Statculator. (2026, November 14). T Critical Value Calculator. Retrieved November 14, 2026, from https://statculator.com/p-values-critical-values/t-critical-value-calculator/