We use a non‐parametric approach to investigate the (inverse) relationship between farm productivity and farm size. A kernel regression is used on data of mixed cropping systems to study the determinants of production including different factors that have been identified in literature as missing variables in the testing of the inverse relationship such as soil quality, location and household heterogeneity. Household data on farm activities and crop production were gathered from 640 households in 2007 in two Northern provinces of Burundi. Our results do not reject the findings of an inverse relationship between farm size and productivity. However, we find that size returns vary substantially with farm size, that is, between 0.2 for the smallest farms and 0.8 for the largest farms. Other factors that significantly affect production include soil quality. Finally, we find a significant positive association between food security and farm size.