Although it is well known theoretically that physicians respond to financial incentives,the empirical evidence is quite mixed.We use the 2004 Canadian National Physician Survey to obtain the number of patient visits per week to physicians office as a measure of output,to analyze the behaviour of family physicians in alternative forms of remuneration schemes.We find that family physicians self-select into different remuneration regimes based on their personal preferences and unobserved characteristics.Moreover,we use OLS estimates plus the estimates from an Ⅳ GMM procedure to tease out the magnitude of the selection effect.For instance,compared to fee-for-service,salaried physicians would conduct about 18％ more patient visits per week as a result of the selection effect,while they would conduct about 54％ fewer patient visits per week as a result of the pure incentive effect.Knowledge of how physicians self-select into remuneration schemes and how these schemes affect physician output are important considerations for effective health policy formulation.