The existence of strong spill-over effects from private R&D increases the potential social contribution of R&D, but may depress the private incentives to undertake it. R&D consortia offer a potentially effective means of internalising this externality. Governments in Europe and North America have adopted policies to promote R&D consortia formation, motivated less by economic theory than by the perception that the Japanese government has used such policies to great effect (Tyson 1992). Despite the existence of a large theoretical literature analysing the potential benefits and costs of R&D consortia, there has been little corresponding empirical work on their efficacy. The authors undertake the first large-sample econometric study of Japanese government-sponsored research consortia, using firm-level data on research inputs and outputs to measure the impact of participation on the ex-post research productivity of the firm. Analysis shows that frequent participation in these consortia has a positive impact on research expenditure and research productivity. These results hold after controlling for the potential endogeneity of the intensity of participation in consortia to participating firms' research productivity. Furthermore, evidence is provided to show that part of this impact arises from the increased knowledge spill-overs that take place within these consortia.