The association between foreign aid and growth has been controversial for decades and the evidence is quitemixed with results sensitive to data samples and modelling approaches. We reassess the relationship betweenaid, policy and growth implementing a novel semiparametric estimation method that allows for nonlineari-ties and control for endogeneity. We use a panel of 61 developing countries for the period 1970–2005. Theresults show that the aid-policy-growth relationship is complex with varying correlation patterns across dif-ferent aid and policy levels. Aid does not seem growth-enhancing, except at very high levels (above 7% ofAid/GDP), whereas the effect may arrive with a lag at lower levels. Good policies are positively associatedwith growth and is more obvious in better policy environments. Aid effectiveness does not increase withbetter policies.