Are there any drawback to using PERT?
The PERT approach, even used as originally intended, has been shown to produce inaccurate and biased numbers. We have known of this problem since the early-1960s (MacCrimmon and Ryavek 1962; Archibald and Villoria 1967; Hulett 2007) though some people apparently did not get the memo. The weakness of the PERT process is that it recognizes only the PERT critical path and does not recognize that parallel slack paths can contribute to risk at the merge points. PERT ignores the contribution of those slack paths and hence underestimates schedule risk at the merge points, the well-known merge bias. However we do not know the extent of this bias without doing a Monte Carlo simulation, so we cannot say that the bias is small or minimal. PERT underestimates the risk for all schedules where there are parallel paths, and this means it underestimates risk for all real schedules. This underestimation occurs at each merge point within the schedule structure. Also See: http://blogs.ittoolbox.com/eai/implementation/archives/pros-cons-of-the-pertcpm-method-22221