In this paper, we present an in‐depth analytical study of a semi‐preemptive priority scheduling discipline. This discipline eliminates the deficits of both the full‐ and non‐preemptive versions. Under the non‐preemptive category, in particular, higher‐priority customers may have to wait even when the service of a lower‐priority customer has just started, while under the full‐preemptive discipline, the almost completed service of a lower‐priority customer may be interrupted due to the arrival of higher‐priority customers, possibly causing a large extra delay. For fixed low‐priority service times, the semi‐preemptive priority scheduling discipline shows a performance gain of up to 6% compared to the full‐ and non‐preemptive versions.
