OBJECTIVE: Making process and technology changes in clinical practice is often complex and costly. The interdependence of processes, staff, technology and patients can bias static models. Dynamic system simulations can identify problems, and solutions, by imitating actual processes over time. Our objective was to examine the efficiency possible with increased resources in an outpatient endoscopy unit using discrete event simulation. METHODS: We designed a computer simulation model (AnylogicTM, AccessTM) of a three procedure room outpatient endoscopy unit with eight shared preparation/recovery rooms. Design involved documenting patient and staff flow; determining process times via observation studies, chart reviews and staff interviews; and validating output with actual performance data. The model simulates 20 working days and generates mean daily statistics. Outcome measures include patient throughput (patients per day, PPD) and length of stay in minutes (LOS), calculated as mean patient time from arrival to discharge. Scenarios included doubling staff, adding combination responsibility nurses or adding prep/recovery rooms to the baseline model. RESULTS: At baseline, the unit averaged 17 PPD, 177 minute LOS. Outcome statistics for scenarios are (PPD, LOS): doubled staff (18, 164), one nurse (18, 168), two nurses (18, 166), one room (19, 184), two rooms (20, 192). Two additional prep/recovery rooms results in the largest throughput impact (+3 PPD), but increases LOS 15 minutes. The addition of one nurse results in 98% of the throughput achieved with doubled staff (+1 PPD) and reduces LOS by nine minutes. CONCLUSIONS: Increasing space capacity results in the largest throughput in this unit suggesting a potential solution to patient backlogs over time. However, adding one nurse both increases productivity and reduces patient LOS. Diminishing marginal returns for increased nursing units were observed. Further modeling may illuminate strategies to optimize efficiency and profit while also improving measures of customer satisfaction (i.e., LOS).