Policymakers all over the world, but particularly in developing countries, are struggling with the claim that too little is being invested in public agricultural research and development (R&D). This briefing paper tries to clarify the issue by introducing a simple, stylized model is based on the concept of an ex ante choice set of all conceivable agricultural R&D projects, which, when ranked by their expected rate of return (ERR), form a distribution that increases steadily (and, we assume, exponentially) with decreasing ERR. The economic selection of R&D projects follows the simple optimizing rule of always selecting the project with the highest ERR first, then the next highest, and so on, until the budget is exhausted or until the ERR of the last selected R&D project equals the social rate of return, whichever comes first. The economically optimal investment level is to finance all R&D projects with an ERR that is equal to, or higher than, the social rate. This simple model highlights two distinctive aspects of the underinvestment problem: (1) suboptimality in project selection-some projects below the optimal cutoff point are selected at the expense of projects above the optimal cutoff point, and (2) the underlying factors that shape the set of agricultural R&D investment opportunities from which to choose. A better understanding of these factors, and those that cause selection suboptimality, may provide important insights into the variables that could help pull (rather than push) additional resources into agricultural R&D.