Journal of Research in Medical Sciences1735-1995622001060234643464ENBiostatistics and Epidemiology Department, School of Health, Isfahan University of Medical Sciences and Health Services, Isfahan2001060220090412In many casses the estimation of variance which is used to determine sample size in clinical trials, derives from limited primary or pilot studies in which number of samples is small. since in such casses the estimation of variance may be much far from the real variance, the size of samples is suspected to be less or more than what is really needed. In this article an attempt has been made to give a solution to this problem. in the case of normal distribution. Based on distribution of (n-1) S2/?2 which is chi-square for normal variables, an appropriate estimation of variance is determined an used to calculate sample size. Also, total probability to ensure specific precision and power has been achived. In method presented here, The probability for getting desired precision and power is more than that of usual method, but results of two methods get closer when sample size increases in primary studies.http://jrms.mui.ac.ir/index.php/jrms/article/view/3464