A parallelism analysis integrated system (PARIS) with a multiple instruction stream-multiple data stream (MIMD) scheme has been developed to analyze simulation programs and generate a parallel execution program for parallel processing. This simulation program can predict effects of anomalies in nuclear plants. The PARIS system first analyzes task parallelism and the processing time of each task after a user divides a program developed for a single processor into many elementary assignment units. The system then assigns tasks to processors using the critical path/most immediate successor first scheduling algorithm to minimize the overall processing time, and it generates the parallel execution program, which can be executed with a tightly coupled multiprocessor. The PARIS system has two scheduling methods so it can assign tasks to the multiprocessor both before and during execution of the program. Thus, optimum task scheduling is accomplished even when the processing time of each task changes according to accident analyses. The PARIS system is assessed using a nuclear power plant analyzer code (NUPAC-1) that includes reactor coolant system and steam generator models. The results show that the NUPAC-1 processing time with 7 processors is 3.5 times as fast as with a single processor. The fast-running capability is 5.4 times as fast as real time in steady-state and transient analyses and 4.0 times as fast in accident analyses. Furthermore, the results show that the PARIS system can be adapted to realize a predictive simulator using the NUPAC-1 code with few nodes.