Purpose: This paper introduces a procedure for solving the staffing problem in a service system (i.e., determining the number of servers for each staffing period). Design/methodology: The proposed algorithm combines the use of queueing theory to find an initial solution with the use of simulation to adjust the number of servers to meet previously specified target non-delay probabilities. The basic idea of the simulation phase of the procedure is to successively fix the number of servers from the first staffing period to the last, without backtracking. Findings: Under the assumptions that the number of servers is not upper-bounded and there are no abandonments and, therefore, no retrials, the procedure converges in a finite number of iterations, regardless of the distributions of arrivals and services, and requires a reasonable amount of computing time. Originality / value: The new procedure proposed in this paper is a systematic, robust way to find a good solution to a relevant problem in the field of service management and it is very easy to implement using no more than commonly accessible tools.