Abstract: The impact of jobs random fluctuation on scheduling rule is analyzed, the processing time of which subjects to exponential distribution. For two jobs and three jobs, with theory of probability and statistics, analytical solution of failure probability are obtained. For multiple jobs, by proposing a simulation of random number generator model, and presenting the unbiased estimator and confidence interval, the iteration times of simulation and effectiveness of the model are analyzed. Finally, the following conclusion is obtained that for jobs with exponentially distributed processing time, the failure probability gets higher when ratios of every two adjacent jobs get closer to 1. The research affords a theoretical guidance for job scheduling.