The objectives of this work are to model a manufacturing system using top-down Petri net modeling approaches and mutual exclusion concepts; to construct stochastic Petri nets by considering the random failures of the machines, as well as such resources as robots and automated guided vehicle (AGV) systems, and by incorporating temporal variables to transitions and places; to derive the performance of the system using stochastic Petri net performance models for different cases; and to compare performance results. Performance analysis problems on both deadlock-free and deadlock-prone systems are addressed and a comparison between them is made. First the stochastic Petri net modeling process is discussed on the basis of top-down and bottom-up ideas. Then, system performance indices, such as throughput for a resource-sharing manufacturing system, are derived by using existing software packages such as SPNP. Finally, conclusions are drawn and future directions are discussed for Petri net evaluation of manufacturing systems.