Extending the analysis outlined in “Not Your Average Job: Measuring Farm Labor in Tanzania,” this paper explores the implications of our findings on recall bias for assessing both the extent of the productivity gap in agriculture, and the possibility of allocating rural labor more efficiently across sectors—for instance, into other (non-agricultural) income-generating activities. We find that the agricultural productivity of recall-surveyed households is 78% lower than that of weekly-surveyed households. Decompositions of the mean difference in productivity indicate that the number of hours of farm labor reported by the households explains a large portion of this difference. Given our finding that recall households substantially over-report hours worked while also under-reporting key inputs such as plots, these results suggest that the measures of labor productivity obtained from traditional recall surveys may be misleadingly low. Such problems in labor measurement could thus distort our view of the true productive capacity of small-holder farms and of the optimal allocation of rural labor. In ongoing analysis, we test for similar differences in non-agricultural labor productivity, and we reassess the contribution of human capital to labor productivity in agriculture in light of the inverse relationship we find between recall bias and the level of education.