More and more emphasis is allocated to production job scheduling by the modern organizations in order to preserve their performance indicators stable. This research addresses the dynamic scheduling of jobs in a production system with a single input queue and parallel machines. The processing times and the times between job arrivals are assumed probabilistic. Jobs belong to different classes and a due date is assigned to each job at the time of its arrival. A machine needs to be setup every time it switches production from some job class to another. In this article, is considered a set of alternative priority rules for dynamic job scheduling using discrete-event simulation. The priority heuristics are compared in respect to several performance metrics in a series of simulation experiments. The behaviour of the scheduling heuristics is assessed under the influence of various parameters. Moreover, are offered managerial insights for scheduling decisions in industry based on the numerical results.