Aiming at the practical conditions of stochastic flow network system, a combined method for estimating the reliability based on rough sets theory and Petri nets is proposed. The Petri net model for the corresponding weighted graph of a certain network state is constructed. Then the importance indexes of each arc states to the system state are obtained from the knowledge base by applying the rough sets theory, which can be regarded as the priority index of transitions in the Petri net model. Therefore, the firing of transitions in the model can be controlled. Finally, the Monte-Carlo method is used to obtain the estimated reliability of stochastic flow network system. Simulation results show that the combined method is efficient to calculate the reliability of stochastic flow network system.