In 2009, a system of pay-for-performance (P4P) was offered to physicians in France via the Contrat d’Amélioration des Pratiques Individuelles (CAPI). This study assesses the causal impact of CAPI on their behaviour in terms of care provision. Based on a panel of general practitioners in private practice observed before (2005 and 2008) and after (2011) its introduction, we use an instrumental variables approach, applied to a model in first-differences in order to correct the endogeneity biases linked to the fact that signing up to CAPI is a choice. We show that, unlike other practitioners, those who have signed up to CAPI have not reduced their number of consultations per patient or the amount of prescriptions per patient. They have also increased, to a greater extent than others, the proportion of their patients who they treat as the primary care doctor(i.e. the médecin traitant). Moreover, CAPI has enabled them to increase their fees per patient with, as a consequence, a higher treatment cost for the Social Security system.