Our objective was to implement and evaluate performance-based reimbursement for influenza immunization of the elderly in physician offices. We performed a community-based quasi-experiment with historic and concurrent comparisons, using primary care physician offices in Monroe County, New York. Participants in the intervention group included 53 primary care physicians admitting to one hospital, and the comparison group included 82 primary care physicians admitting to other hospitals. All physicians participated in a Medicare-sponsored demonstration to increase influenza immunization rates, and, during the 1990-1991 immunization season, used a target-based poster to track immunization rates. Physicians in the intervention group were enrolled in a performance-based financial incentive program that rewarded immunization rates above 70%. A survey concerning influenza immunization practices and opinions was sent to all physicians. The average physician-specific immunization rate in the incentive group was 73.1% versus 55.7% in the comparison practices (P < .001). Eligibility for incentives, practice size, sex of physician, medical specialty, reminder postcards, and practice populations including medically indigent patients were associated with immunization level. Controlling for the above variables, we completed a regression analysis showing that eligibility for the incentive was still significant (P = .003). The survey responses were not predictive of performance or significantly different between the two groups, except for the negative influence of sending postcards. This study in a community setting suggests that linking reimbursement to performance may be a successful strategy to increase influenza immunization levels for the elderly.