In many queuing situations customers are served in order of arrival; yet if customers with shorter service times were served first, it is well known that the mean queuing time could be reduced considerably. In a comparison with the ″optimal″ shortest service-time rule, a simple queue discipline involving only two non-preemptive priority classes fares very well. Numerical results illustrate the practical effects of implementing this simple alternative to service in order of arrival in a variety of queuing situations.