The Program Evaluation and Review Technique (PERT) has become a classic Project Management tool for estimating project duration when the activities have uncertain durations. However, despite its simplicity and widespread adoption, the original PERT, in neglecting the merge event bias, significantly underestimated the duration average and overestimated the duration variance of real-life projects. To avoid these and other shortcomings, many authors have worked over the last 60 years at producing interesting alternative PERT extensions. This paper proposes joining the most relevant of those to create a new reformulated PERT, named M-PERT. M-PERT is quite accurate when estimating real project duration, while also allowing for a number of interesting network modelling features the original PERT lacked: probabilistic alternative paths, activity self-loops, minima of activity sets and correlation between activities. However, unlike similar scheduling methods, M-PERT allows manual calculation through a recursive merging procedure that downsizes the network until the last standing activity represents the whole (or remaining) project duration. Hence, M-PERT constitutes an attractive tool for teaching scheduling basics to engineering students in a more intuitive way, with or without the assistance of computer-based simulations or software. One full case study will also be proposed and future research paths suggested.