In this study, a two-stage multi-objective surgery scheduling method for a multiple operating rooms (ORs) system is developed. Minimizing the total surgery operation cost (overwork and surgery cancel cost) and OR set up cost are considered as the ultimate objectives in this method. Surgery priorities are evaluated based on patient age, surgery types, estimated surgery durations, and delayed days in the first stage using regression technique. In the second stage, the proposed priority-based surgery scheduling optimization model considering OR specialties is solved using the discrete harmony search (DHS) algorithm. Daily-based experiments are performed and analyzed using simulated data sets with 20 and 100 medical records, respectively. Experimental results show that the near optimal surgery schedules can be obtained efficiently and accurately from the proposed twp-stage surgery scheduling method.