Virtually all models,of the household,have the minimal,implication that the equilibrium allocation of resources is Pareto efficient. Within many African households, agricultural production is simultaneously carried out on many plots controlled by different members,of the household. Pareto efficiency implies that variable factors should be allocated efficiently across these plots. This paper provides a simple test of this weak implication of household models,using an extremely detailed agronomic,panel data set from Burkina Faso. I find that plots controlled by women,have significantly lower yields than similar plots within the household planted with the same crop in the same year, but controlled by men. The yield differential is attributable to significantlyhigher labor and fertilizer inputs per acre on plots controlled by men. These results contradict the Pareto efficiency of resource allocation within the household. Production function estimates imply that about six percent of output is lost due to the misallocation of variable factors across plots within the household. The paper concludes with suggestions for a new model of intra-household allocations consistent with the empirical results. , Samuelson (1956) and Becker (1991) provide alternative conditions under which households act as if they ,