Petri nets represent a useful tool for performance, dependability, and perf
ormability analysis of complex systems. Their modeling power can be increas
ed even more if nonexponentially distributed events are considered. However
, the inclusion of nonexponential distributions destroys the memoryless pro
perty and requires to specify how the marking process is conditioned upon i
ts past history. In this paper, we consider, in particular, the class of st
ochastic Petri nets whose marking process can be mapped into a Markov regen
erative process. An adequate mathematical framework is developed to deal wi
th the considered class of Markov Regenerative Stochastic Petri Nets (MRSPN
). An unified approach for the solution of MRSPNs where different preemptio
n policies can be defined in the same model is presented. The solution is p
rovided both in steady-state and in transient condition. An example conclud
es the paper.