Objective: The purpose of this study was to determine stability of physician performance rates on process metrics by patient volume threshold level and by alternative algorithms of deriving a global physician performance score. Methods: Preferred provider organization (PPO) health plan claims data between April 2003 and March 2005 were analyzed. Six alternative algorithms of calculating a global metric rate from several process metrics were compared. Stability was also assessed across the different patient thresholds within each algorithm, where patient volume thresholds of 0, 5, 10, 15 and 20 were applied for each process measure. Trend test was used to compare longitudinal stability across patient volume thresholds and across global score algorithms. Results: A total of 2036 generalist and specialist physicians were included. The sample size of eligible physicians decreased with increasing minimum patient volume thresholds. The greatest drop occurred when the threshold was increased from 0 to 5 patients (17-21% decrease, varying by year). The average indicator denominator size ranged from 95 patients (S.D.±160) in 2003 to 172 patients (S.D. ±337) in 2005. Physician performance rates were least stable when no minimum patient volume threshold was required for analysis. No significant differences in score stability over time were observed between the different patient volume thresholds of 5 patients or above. Conclusion: A minimum patient volume threshold of 5 patients is likely necessary to preserve stability of physician performance rates over time. Increasing the level of threshold beyond 5 patients, however, did not seem to significantly increase longitudinal stability of performance rates.