In a recent paper, Delgado and McIntire argue that lack of profitability is a sufficient explanation for nonadoption of animal traction using oxen (ATO) in the Sahel. The purpose of this comment is to present alternative explanations for the nonadoption of ATO and to debate some of the points made by Delgado and McIntire concerning the labor requirements of ATO. The magnitude of the ATO investment and the long learning period involved in perfecting its use both argue for a multiyear perspective in assessing profitability rather than the single-year perspective used by Delgado and McIntire.' Oxen traction is not a simple-or short-term investment for the peasant farmer. Experience with a recent ATO project in eastern Upper Volta (Barrett et al.) indicates that a fourto six-year learning period is required for a new adopter to realize the full technical benefits of ATO. Eight years may elapse before the investment breaks even. Because of high initial investment costs and deferred benefits, the long-run profitability of ATO, as measured by net present value or internal rate of return, is generally much lower than that indicated by single-year analysis (Lassiter). Even where long-run ATO profitability is expected, severe cash flow problems are likely in the short run. Ten-year cost and return projections for eastern Upper Volta suggest a 20% to 30% income reduction for ATO adopters during the first four years of the investment, despite an internal rate of return of 15% (Lassiter, p. 233). Because many Sahelian farmers sell little of their farm output, covering such cash flow deficits would require either (a) nonfarm cash income or (b) a shift toward cash crops, which could jeopardize family food supplies. A sample of 125 relatively successful ATO adopters in Upper Volta contained substantially larger households with proportionally more nonfarm cash income than nonadopters (Barrett et al.). Furthermore, the extremely variable climate of the Sahel confronts adopters of ATO with major financial and technical risks which long-run average profitability indicators hide. By adversely affecting yields and income, poor weather could worsen cash flow deficits critically, especially in years soon after adoption. Lack of institutional support also discourages adoption of ATO. Successful use of ATO requires a comprehensive network of services, such as agricultural extension, marketing, animal training, veterinary care, credit, and equipment repair. However, provision of these services on an integrated basis has been characteristically weak in most ATO projects (Sargent et al.). Also the majority of projects has been small and underfunded, and most Sahelian farmers have not had an opportunity to adopt ATO. Thus, the low adoption of ATO in the Sahel is partly a result of institutional failure rather than a widespread decision by farmers not to adopt the technology, as Delgado-McIntire imply. The Delgado-McIntire paper emphasizes the important influence of the opportunity cost of labor on the farmer's decision regarding ATO adoption. A shortcoming of their analysis is that it exaggerates the quantity, timing, and quality of labor required to maintain draft oxen. Their core linear programming (LP) coefficients (pp. 191-92) apparently are based on Delgado's original Tenkodogo study, which used labor data for herding range cattle as a proxy for the maintenance requirements of draft oxen. Thus the Delgado-McIntire model assumes that a full-time adult male is required to care for two draft oxen uring the growing season, as would be the case for range cattle. This clearly attaches a prohibitive labor cost to ATO and seriously constrains production possibilities in their LP model. Draft oxen do not require such intensive surveillance. Unlike wild range cattle, draft oxen are trained, manageable animals which can be cared for by young children of either sex. Because oxen are stall-fed, labor requirements are minimal except when forage crops are harvested. Data on twelve pairs of draft oxen in eastern Upper Volta (Lassiter, pp. -173-77) indicate that the monthly worker equivalent'2 labor requirements for animal mainteEric W. Crawford is an associate professor, Department of Agricultural Economics, Michigan State University. Gregory C. Lassiter is a professor, Department of Agricultural Economics and Education, California State University, Fresno. The authors gratefully acknowledge the helpful comments of the anonymous reviewers. Review was coordinated by Hans P. Binswanger, associate editor.