The study establishes the link between agricultural labour productivity and natural resources at the national level in Honduras. Using ordinary least square regression and GIS, we identified why some villages enjoy higher labour productivity than others. For every village in Honduras, we constructed 44 natural resources and socioeconomic variables that could have an impact on productivity. Soil variables were obtained by interpolating, within geological regions, the results of the analysis of 4000 soil samples. Climate variables were obtained by interpolating long-term averages of monthly data from 400 meteorological stations. The 25 socioeconomic variables, most of them derived from unit-level census data, were chosen to describe the main farm characteristics and their access to markets. We show that the impact of natural resource conditions on agricultural productivity is not as direct as could be imagined. Soil and terrain slope and altitude had little impact because the valuable coffee production, which is located in the hillsides, compensates for the supposed handicap of hillside areas. The length of the rainy season, however, showed the strongest correlation with labour productivity. Socioeconomic factors such as population density, adoption of new germ plasm, technical assistance and education were found to be strong determinants of labour productivity. The conclusion is that rainfall is the main factor. Thus migrations that are currently occurring from the dry, highly populated areas towards the rainy, less populated ones will improve the situation. There is also great scope to improve labour productivity through proper public investment.