OBJECTIVE:To determine whether area-level Medicare physician fees for mastectomy and breast conserving surgery were associated with treatment received by Medicare beneficiaries with localized breast cancer and to compare these results with an earlier analysis conducted using small areas (three-digit zip codes) as the unit of observation. DATA SOURCE:Medicare claims and physician survey data for a national sample of elderly (aged 67 or older) Medicare beneficiaries with localized breast cancer treated in 1994 (unweighted n = 1,787). STUDY DESIGN:Multinomial logistic regression analysis was used to estimate a model of treatment received as a function of Medicare fees, controlling for other area economic factors, patient demographic and clinical characteristics, physician experience, and region. PRINCIPAL FINDINGS:In 1994, average Medicare fees (adjusted for the effects of modifiers and procedure mix) for mastectomy (MST) and breast conserving surgery (BCS) were 904 dollars and 305 dollars, respectively. Holding other fees and factors fixed, a 10 percent increase in the BCS fee increased the odds of breast conserving surgery with radiation therapy relative to mastectomy to 1.34 (p = 0.02), while a 10 percent decrease in the MST fee increased the odds of breast conserving surgery with radiation therapy to 1.86 (p < 0.01). CONCLUSIONS:Among older women with localized breast cancer, financial incentives appear to influence the use of mastectomy and breast conserving surgery with radiation therapy. This finding is consistent with the hypothesis that physicians are responsive to financial incentives when the alternative procedures have clinically equivalent outcomes and the patient's clinical condition does not dominate the treatment choice. We also find that the fee effects derived from this analysis of individual data with more precise measurement of both diagnosis and treatment are qualitatively similar to the results of the small-area analysis. This suggests that the earlier study was not severely affected by ecological bias or other data limitations inherent in Medicare claims data.