Panel data from Rwanda allow us to explore costs and benefits from land fragmentation in a non-mechanized setting using two methodological improvements, namely (i) a terrain-adjusted measure of travel time/cost required to visit all parcels to measure fragmentation; and (ii) instrumental variable (IV) approaches that use measures for inherited/allocated parcels and past displacement as instruments. Results suggest that fragmentation as measured by travel cost negatively affect yield, intensity of labor use, and technical efficiency while reducing yield variability. With some 7 percent increase in yields, the size of the estimated impact of potential consolidation remains modest, suggesting that in an unmechanized setting such as the one studied here, the costs of programs to reduce fragmentation may outweigh the benefits.