Pay-for-performance programs are being developed to improve quality of care despite limited empirical evidence demonstrating their effectiveness, especially in underserved communities. Using data on 1,166 patients treated by 46 primary care physicians, this paper examines the effect of an innovative pay-for-performance program implemented in 2004 at the nation's largest federally qualified health center on hemoglobin A1c (HbA1c) testing and HbA1c scores. Univariate analysis and logistic regression results show that the pay-for-performance program significantly increased the likelihood that patients received two HbA1c tests per year as recommended by the American Diabetes Association. Ordinary least squares (OLS) regression results reveal that physicians' baseline performance was positively related to performance after program implementation. Finally, OLS regression results suggest that the program did not contribute to improved blood sugar control. The pay-for-performance program appeared to improve compliance with HbA1c testing recommendations, but a more comprehensive strategy, including increased patient support, may be necessary to improve health outcomes in disadvantaged populations.