The ta-test is a modified two-sample or two-group t-test of Gosset (1908). In small samples with less than 15 replicates,the ta-test significantly reduces type I error rate but has almost the same power with the t-test and hence can greatly enhance reliability or reproducibility of discoveries in biology and medicine. The ta-test can test single null hypothesis or multiple null hypotheses without needing to correct p-values.