Food security in Tanzania is projected to deteriorate as a result of climate change. In spite of the Government’s efforts to promote agricultural practices that improve productivity and food security, adoption rates of such practices remain low. Developing a thorough understanding of the determinants of adoption and updating our understanding of the impacts of these technologies under the site-specific effects of climate change are crucial to improve adoption. This paper addresses these issues by using a novel data set that combines information from two largescale household surveys with geo-referenced historical rainfall and temperature data in order to understand the determinants of the adoption of sustainable and productivity improving practices and their impacts on maize productivity in Tanzania. The specific practices analyzed are: maize-legume intercropping, soil and water conservation practices, the use of organic fertilizers, inorganic fertilizers and high yielding maize varieties. We find that farmers located in areas where the cropping season’s rainfall has been highly variable have 15 percent lower maize yields. Similarly, farmers located in areas where maximum temperatures during the growing season exceed 30 degrees Celsius have approximately 25 percent lower yields. Both rainfall variability and hotter temperatures are expected to increase under climate change, underlining the importance of policies to buffer food security from the estimated effects of climate change. Our analysis identifies policy entry points both to improve maize productivity and the adoption of practices to do so through careful empirical analysis. This paper contributes to evidence base to support policies for climate smart agriculture and underlines the importance of integrating site-specific analyses of climatic variables in policy targeting to foster adoption of appropriate practices to improve food security under climate change.