This paper presents a conceptual analysis of important issues in management of agricultural research drawing on principal-agent theory and derives implications for funding and management of agricultural research. Building on well known attributes of research, whereby research results are risky, outputs are uncertain and sometimes unanticipated, more than one approach has validity for a given topic, we consider how incentives should be structured to elicit optimal research effort and directions, whether research directions should be set at a centralized or decentralized level, and the optimal duplication of effort. The results suggest that (i) the current trend toward replacement of formula funding by competitive grants allocation may be ill conceived, (ii) a mixed system with some research funding and direction at the federal level, some at the state level, and perhaps some at regional levels is advantageous, and (iii) funding of competing scientists working on the same problem at different institutions has merit.