The purpose of the paper is to theoretically examine the welfare implications of public sector involvement in agricultural biotechnology R&D. The model assumes that firms (either a private duopoly consisting of a pair of for-profit firms or a mixed duopoly consisting of one for-profit firm and one public firm) compete in a winner-take-all patent race that is subject to R&D spillovers. Unlike previous research, spillovers are explicitly incorporated into the race, and the size of the prize that accrues to the winner, as well as the size of the ex post social surplus, is contingent on whether or not the public firm participates in the stage two product market. The welfare results concerning the implication of public sector involvement in the R&D race have several unexpected properties because of the interaction of the three sources of market failure (underinvestment due to spillovers, overinvestment due to winner-take-all racing and monopoly pricing in the product market). The main result is that the R&D subsidy or tax, while effective at improving the efficiency of the R&D outcome, is not effective at correcting the monopoly pricing market failure in the product market. The public firm, on the other hand, is able to simultaneously shift the R&D outcome toward first-best and reduce the expected distortion in the product market. The public firm invests particularly aggressively when R&D spillovers are high and the deadweight loss from monopoly pricing in the product market is high. An important problem with public firm participation in the R&D race is that cost smoothing inefficiencies arise because the public firm will either invest at a relatively high level to address the underinvestment externality, or invest at a relatively low level to address the overinvestment externality. Cost smoothing considerations always prevents attainment of first-best when product market externalities are present.