This paper empirically analyses how a public institution chooses the cofinancing rate in funding competitive agricultural R&D research projects. The public funding institution observes some objective features of the selected research projects and of the proponents. The paper puts forward some testable hypotheses about how the funding institution uses this available information to decide the cofinancing rate. An empirical model is then specified and estimated to test these hypotheses. The empirical application refers to the real case of the agricultural R&D program funded by an Italian region (Emilia-Romagna) over years 2001-2006. Results suggest that the cofinancing rate actually responds to the observed features but this response is not always consistent with the formulated hypotheses.