Various cases of unequal variances and unequal sample sizes from a normal and a skewed population were used to empirically obtain the probability of a Type I error and the power for the permutation t-test as compared to Student's t-test and the Mann-Whitney U-test. Empirical results showed differences for different sample sizes, variance ratios, population sampled, and size of mean of the population. The power of the permutation t-test is very close to or greater than that of Student's t-test for both populations, and the power is large if the large variance accompanied the large mean for the skewed population. (Author)