In discrete event system simulation, random samples from probability distributions are typically used as input for a model. Since these samples are just a sequence of observations generated by random processes that could have large variability, therefore the results of each run can vary substantially. As a consequence, a suitable number of replications must be executed in order to provide good accuracy on estimation of key performance indicators. This paper aims to propose two systematic approaches to manage simulation campaigns automatically. The method behind these approaches increase the runs’ number iteratively in order to reduce error on performance indicator estimation, thus respecting precision requirements on simulation output results. The presented approaches can be used for both terminating and non-terminating systems.