This paper develops a simple model to study research joint ventures (RJV) for product innovation with uncertainty. Equilibrium RJV structure, members' expected profits and expected time for successful product innovation are analyzed in our framework. The results demonstrate that the equilibrium RJV structure is non-coalition if the competition effect in the product market is relatively large. Complementary coalition is obtained if there exists better synergy of complementary RJV. Grant coalition is shown to be the equilibrium when the scale effect of homogenous RJV is relatively large. Our study also suggests that complementary coalition provides earlier product innovation than others if the synergy is relatively large. Otherwise grant coalition provides earlier innovation.