Critical values

T critical value calculator

Find the critical t (often written t*) for any degrees of freedom and alpha. You get the two-tailed value, the one-tailed value, and the multiplier for the matching confidence interval, with R's qt and SciPy's t.ppf code.

  • Two-tailed and one-tailed values
  • Any df, including decimals
  • The t* for confidence intervals

Sources checked on November 21, 2026. Results computed by the site's engine and tested against R and SciPy.

Critical t for alpha = 0.05 with 10 df

Two-tailed±2.2281
One-tailed, upper1.8125
One-tailed, lower−1.8125
−4−2024

A two-tailed test at alpha = 0.05 rejects when |t| is above 2.2281; this is also the multiplier for a 95% confidence interval. The one-tailed value 1.8125 puts all of alpha in one tail.

Reproduce this result in R or Python

R and Python give the two-tailed value; use 1 minus alpha instead of 1 minus alpha/2 for one tail.

R

qt(0.975, df = 10)

Python (SciPy)

from scipy.stats import t
print(t.ppf(0.975, 10))

What a critical value is

The critical value is the point on the t distribution beyond which a test rejects the null hypothesis. For a two-tailed test at alpha = 0.05 it is the t that leaves 2.5% in each tail; for a one-tailed test at the same alpha it leaves the full 5% in one tail. The same number is the multiplier in a confidence interval: a 95% interval is the estimate plus or minus t* times the standard error, with t* the two-tailed 5% critical value.

Worked example: 10 degrees of freedom

With df = 10 and alpha = 0.05, the two-tailed critical value is 2.2281 and the one-tailed value is 1.8125. NIST's table prints 2.228 and 1.812, the same numbers rounded to three decimals. A t of 2.5 from a two-sided test with 10 df would pass the 5% line; a t of 2.1 would not.

dfTwo-tailed, alpha 0.10Two-tailed, 0.05Two-tailed, 0.01
16.31412.70663.657
22.9204.3039.925
52.0152.5714.032
101.8122.2283.169
201.7252.0862.845
301.6972.0422.750
601.6712.0002.660
1201.6581.9802.617
infinity1.6451.9602.576

The full grid, with more probabilities and every df from 1 to 120, is in the printable t table.

Why the value shrinks as df grows

With few degrees of freedom the sample SD is a rough estimate of the true SD, so the t distribution has heavy tails and the critical value is large: 12.71 for one df at the two-tailed 5% level. As df grows, the estimate firms up and the critical value falls towards the normal value of 1.960. By 30 df it is 2.042, and by 120 df it is 1.980. The t distribution guide shows the curves.

Decimal degrees of freedom

Welch's test produces df such as 18.33. The calculator accepts them directly; the critical value for that df at the two-tailed 5% level is 2.0982, which sits between the values for 18 and 19 df. If you are using a printed table instead, rounding the df down gives a slightly larger, more cautious critical value.

Confidence levels and alpha

Confidence levelTwo-tailed alphaCritical t, df = 10
90%0.101.8125
95%0.052.2281
99%0.013.1693
99.9%0.0014.5869

Pick alpha before looking at the data. Using 0.05 is a convention, not a law; some fields use 0.01 or stricter. For a one-tailed test, enter alpha as usual and read the one-tailed value.

Comparing your t with the critical value

For a two-tailed test, reject when |t| is larger than the two-tailed critical value. For an upper one-tailed test, reject when t is above the one-tailed value; for a lower one-tailed test, when t is below its negative. Software gives an exact p-value instead, which carries the same decision and more information, and you can get one from the p-value from t calculator.

Sources