Statistics Reference

T-Table for α = 0.01 (t Critical Values)

The strict t-table column: critical values at the 1% level for every degree of freedom from 1 to 100 (plus the large-sample tail). The two-tail column builds 99% confidence intervals; the one-tail column serves stringent directional tests and multiplicity-guarded comparisons.

How to Read This Table

Rows are degrees of freedom — every integer from 1 to 100, then 120, 150, 200, 500, and 1000, finer than the stepped rows of the main table. The one-tail column puts all of α = 0.01 in a single tail (directional tests); the two-tail column splits it across both tails (non-directional tests and confidence intervals). The distribution is symmetric, so lower-tail tests use the negative of the printed value.

99% confidence interval from a sample of 26 (df = 25):

Row df = 25, two-tail column: 2.787.
Interval: sample mean ± 2.787 standard errors — about 35% wider than the 95% version (2.060).
The one-tail column's 2.485 would serve a directional test at 1%.

Common uses of this level:

99% confidence intervals for a mean (two-tail column)
Stringent directional t-tests at α = 0.01
Conservative per-comparison levels under multiple testing

t Critical Values, α = 0.01, df 1–1000

Student's t critical values at significance level 0.01 for degrees of freedom 1 to 1000
df	One-tail α = 0.01	Two-tail α = 0.01
1	31.821	63.657
2	6.965	9.925
3	4.541	5.841
4	3.747	4.604
5	3.365	4.032
6	3.143	3.707
7	2.998	3.499
8	2.896	3.355
9	2.821	3.250
10	2.764	3.169
11	2.718	3.106
12	2.681	3.055
13	2.650	3.012
14	2.624	2.977
15	2.602	2.947
16	2.583	2.921
17	2.567	2.898
18	2.552	2.878
19	2.539	2.861
20	2.528	2.845
21	2.518	2.831
22	2.508	2.819
23	2.500	2.807
24	2.492	2.797
25	2.485	2.787
26	2.479	2.779
27	2.473	2.771
28	2.467	2.763
29	2.462	2.756
30	2.457	2.750
31	2.453	2.744
32	2.449	2.738
33	2.445	2.733
34	2.441	2.728
35	2.438	2.724
36	2.434	2.719
37	2.431	2.715
38	2.429	2.712
39	2.426	2.708
40	2.423	2.704
41	2.421	2.701
42	2.418	2.698
43	2.416	2.695
44	2.414	2.692
45	2.412	2.690
46	2.410	2.687
47	2.408	2.685
48	2.407	2.682
49	2.405	2.680
50	2.403	2.678
51	2.402	2.676
52	2.400	2.674
53	2.399	2.672
54	2.397	2.670
55	2.396	2.668
56	2.395	2.667
57	2.394	2.665
58	2.392	2.663
59	2.391	2.662
60	2.390	2.660
61	2.389	2.659
62	2.388	2.657
63	2.387	2.656
64	2.386	2.655
65	2.385	2.654
66	2.384	2.652
67	2.383	2.651
68	2.382	2.650
69	2.382	2.649
70	2.381	2.648
71	2.380	2.647
72	2.379	2.646
73	2.379	2.645
74	2.378	2.644
75	2.377	2.643
76	2.376	2.642
77	2.376	2.641
78	2.375	2.640
79	2.374	2.640
80	2.374	2.639
81	2.373	2.638
82	2.373	2.637
83	2.372	2.636
84	2.372	2.636
85	2.371	2.635
86	2.370	2.634
87	2.370	2.634
88	2.369	2.633
89	2.369	2.632
90	2.368	2.632
91	2.368	2.631
92	2.368	2.630
93	2.367	2.630
94	2.367	2.629
95	2.366	2.629
96	2.366	2.628
97	2.365	2.627
98	2.365	2.627
99	2.365	2.626
100	2.364	2.626
120	2.358	2.617
150	2.351	2.609
200	2.345	2.601
500	2.334	2.586
1000	2.330	2.581

Other Significance Levels

The t-table overview carries the classic multi-column grid, degrees-of-freedom rules, and the z-convergence walkthrough. The other levels each have a dedicated page:

Frequently Asked Questions

When should t-tests use α = 0.01?

When false positives are expensive (confirmatory research, quality-critical decisions) or as an informal guard when several comparisons run at once. The cost is power: at df = 25, the two-tail cutoff rises from 2.060 to 2.787, so real effects need to be substantially larger to clear it.

How much wider is a 99% interval than a 95% one?

The ratio of the critical values: at df = 25 it is 2.787/2.060 ≈ 1.35, so about 35% wider; at large df it settles near 2.576/1.960 ≈ 1.31. Same data, same center - only the coverage promise and the width change.

Do the 0.01 values also converge to a z value?

Yes: the one-tail column approaches z = 2.326 and the two-tail column z = 2.576 as degrees of freedom grow. The df = 1000 row sits within a few thousandths of those limits, and the convergence is slower than at the 5% level - small samples pay a proportionally larger t-penalty at strict levels.