Pay-for-performance programs offering additional payments to GPs can be used not only to improve the quality of care but also for cost containment purposes. In this paper, we analyse the impact of removing financial incentives in primary care that were aimed at containing hospital expenditure in the Italian region of Emilia Romagna during the period 2002-04. Our analysis draws on regional databanks linking GPs' characteristics to those of their patients (including all sources of public payments made to GPs), together with information on the utilisation of hospital services. We employ a difference-in-difference specification to assess changes in expenditures for avoidable and total hospital admissions. We identify the treatment group with GPs operating in districts where the program is withdrawn during the observation period ("Leavers"). Their performance is compared to that of two separate control groups, namely: GPs working in districts that grant incentives for the entire period ("Stayers"), and those working in districts that never introduced measures for the containment of hospitalisations ("Non Participants"). The comparison between treatment and control groups shows that removing incentives does not result in a worse performance by Leavers compared to both control groups. This supports the policy of removing incentives, as such entail extra payments to GPs which, however, do not seem capable of significantly influencing their behaviour in the desired ways. Our findings complement previous evidence from the same institutional context showing that only those programs that aim to improve disease management for specific conditions - rather than to simply contain expenditure - have proven successful in reducing avoidable admissions for the target population.