In this paper we present measures of land and labor productivity for a group of 98 developed and developing countries using an entirely new data set with annual observations spanning the past three decades. The substantial cross-country and intertemporal variation in productivity in our sample is linked to both natural and economic factors. We extend previous work by dealing with multiple sources of measurement error in conventional agricultural inputs when accounting for observed differences in productivity. In addition to the mix of conventional inputs in agriculture, we find that indicators of quality change in these inputs and the amount of publicly provided infrastructure are significant in explaining cross-sectional differences in productivity patterns.