A combined optimization-simulation model is used to examine the extent of productivity and economic benefits to be gained from adding a physician's assistant to a one-physician ambulatory health care practice. It is shown that previous models (which have predicted productivity gains as large as 90 per cent) have overstated these benefits by ignoring offsetting changes in patients' waiting time, waiting room congestion, practice hours, and supervisory requirements. When all these factors are considered, the productivity gain may be as small as 20 per cent, and the increase in net income may be negligible. The sensitivity of productivity and net income to changes in supervisory requirements is also examined. This paper suggests that a rational explanation exists for the reluctance of physicians to adopt physician's assistants in their practices. The economic benefits derived from adding a physician's assistant may well have been overstated in the existing literature.