The paper investigates the conflicting findings in empirical studies linking land productivity to plot size, livestock ownership, investment in farm assets, and land improvement practices. The conflicting impacts found are partly as a result of different model specifications, the type of data used – panel or cross sectional data – and possibly due to imperfections in rural markets. We control for these problems using household and plot level panel data from rural farmers in Uganda. We find that ownership of cattle has a negative and significant impact on land productivity. Investment in farm related assets, land improvements and other small livestock, however, significantly increases productivity. The conflicting impacts are a result of measurement error. The plot size inverse productivity is robust to different specifications and is largely explained by plot-specific unobserved heterogeneity and imperfections in rural markets.