Productivity and efficiency are key performance indicators of improved seeds. Efficiency differences explain part of the variation in productivity. Improved seeds may affect efficiency because farmers often do not apply inputs at optimum. Improved seeds therefore not only directly affect productivity but also indirectly through efficiency. If productivity and efficiency are not estimated jointly, it creates specification problems and it may (over)underestimate benefits of crop improvement research. Previous studies however estimate the productivity and efficiency impacts of improved seeds independently. In this paper, we estimate the joint impact of improved maize seeds on productivity and efficiency using panel data from maize farmers in Ethiopia. Selection biases associated with seeds choice are addressed by estimating production functions using endogenous switching regressions. Our findings show that improved seeds bring productivity and efficiency gains relative to recycled seeds suggesting that the benefits of improved seeds are underestimated by the amount of productivity (efficiency) gains if either of the two are ignored. Unsurprisingly, improved seeds are more productive than traditional seeds, but tradeoffs between productivity and efficiency exist because farmers are less efficient when they use improved seeds than traditional seeds. Our results may inform policy makers to design strategies that could increase productivity at most efficiency.