A project evaluation and review technique (PERT) is a method for estimating the project completion time. It holds some assumptions in the level of activity and in the level of network of activities. PERT calculates the mean and variance of each activity time based on activity level assumptions and calculates the project completion time according to the computed mean and variance of each activity. This paper presents a new method for estimating the mean and variance of activity time based on two parameters estimated by the expert's knowledge and has no further assumptions about parameters of the Beta distribution. The proposed method is compared with the existing ones based on statistical comparison (i.e., paired t-test & variance analysis). The absolute average error of the new method with a wrong acceptance error is equal to (first kind of error) 1% which is less than the absolute average error of all existing ones in the subjective PERT approximation.