T-test is proposed as a parametric test to evaluate mean of population(s).  Therefore, it is assumed that the data are normally distributed. Recently simulation based research reveal that t-test are robust against non-normality assumption, as long as the distribution of the data are symmetric. For that reason, it is interesting to evaluate the normality test that commonly used before the t-test is applied, particularly on its power and its suitability on symmetrical assumption of t-test on distribution of the data.  On this research, the evaluation on normality test was focused at several tests, which are Anderson Darling, Kolmogorov-Smirnov, Ryan-Joiner, and normality test based on skewness and kurtosis, wether the t-test was restricted in evaluation of mean of one population. The simulation reveals that, in general, Anderson-Darling has greater power than the other test in evaluating the normality of the distribution of the data, wether the normality test based on skewness and kurtosis have greater suitability on symmetrical assumption of t-test on distribution of the data. Hence, it is recommended that the normality test based on skewness and kurtosis are used to evaluate the symmetrical assumption of t-test on distribution of the data.