Chi-squared and Kolmogorov-Smirnov tests are two most widely applied methods for goodness-of-fit test. In application of these tests to hydrological frequency analysis, we often encounter the problem of having data with short record length or sample size. Sample size of a random sample affects the accuracies of parameters estimation and goodness-of-fit test. This study, through stochastic simulation of random variables of normal, log-normal, Extreme Value Type I (EV1), Pearson Type III (PT3) and log-Pearson Type III (LPT3) distributions, compares the performance of and K-S tests with respect to type-I-error and power. When population parameters are known, test outperforms the K-S test with very small type-I-error even for sample size as small as 50. When the goodness-of-fit test is conducted with estimated parameters (the null distribution is specified using estimated parameters), it is found that skewness has significant effect on type-I-error of tests. Power of the test is also found to increase with level of significance, sample size and skewness. Keywords: goodness-of-fit test, stochastic simulation, power, type-I-error