AbstractAs a method of fast simulation of queueing networks by parallel processing there is D‐SSQ which uses the limited‐advance processing scheme. The processing capability of the limited‐advance processing scheme has been shown to increase linearly as the concurrency of the model increases. However, on the other hand, if the updating of the maximum logical time requires considerable time after which no processor backtracks (confirmed time), its performance is degraded considerably. Also, the optimal limit of advance must be estimated from the simulation model. In this paper, to improve on these shortcomings, a time‐event driven scheme is proposed. In this scheme the system clock is managed by a central processor and each processor performs the limited‐advance processing within a certain logical time (clock) of the system time. Second, an adaptive time‐event driven method is presented. In this scheme the optimal limit need not be estimated prior to execution and the global clock is changed dynamically to achieve the maximum performance. The processing capabilities of these two methods are measured by experiments and simulations on a