BACKGROUND: Recent publications have focused on increased operating room (OR) throughput without increasing total OR time. The authors hypothesized that a system of parallel processing for lower extremity joint arthroplasties sustainably reduces nonoperative time and increases throughput. METHODS: The high-throughput parallel processing strategy included neuraxial anesthesia performed in an "induction room" adjacent to the OR, patient selection, an additional circulating nurse, and end-of-case transfer of care to a recovery room nurse who transported the patient from the OR to recovery. Instruments and supplies were prepared in a dedicated sterile setup area. Data were extracted from administrative databases. Group comparisons used standard statistical methods; statistical process control was used to evaluate performance over time. RESULTS: There were 688 historic control cases from 299 days over 16 months, and 905 high-throughput cases from 304 days spanning 24 consecutive months starting September 1, 2004. Throughput increased from 2.6 +/- 0.7 (mean +/- SD) to 3.4 +/- 0.8 arthroplasties per day per room. Nonoperative time decreased by 36 min (or 50%) per case. Operative time also decreased by 14 min (12%) per case. The end time for the high-throughput OR day was only 16 min later than control. Nonoperative time, operative time, and throughput remained significantly improved after 2 yr of operation. Contribution margin increased 19.6%. CONCLUSION: Reorganizing the perioperative work process for total joint replacements sustainably increased OR throughput. Because joint arthroplasties generated a positive margin greater than the incremental cost, the high-throughput system improved financial performance.