Summary: We study the problem of staffing university mathematics support services (MSSs) in which students drop in to the service (without appointment) for tutoring support. Our approach seeks to find the minimum sufficient number of tutors (with appropriate specialities) to present by hour and day to cover student demand with tolerable delays. We employ traditional operational research techniques to aid managers and administrators of MSSs to roster their services. The machine interference type queue is adopted to model the number of student queries within a mathematics support session. We define and solve an appropriate integer program to roster the number of tutors needed to run the service efficiently.