This study aims to contribute to a better understanding of the indigenous management of geographically diverse small-scale agricultural production environments in Africa by investigating the effects of customary land allocation on the use of the land and sustainable agricultural development in Swaziland. This study addresses two questions: (a) to what extent has the heterogeneity of the natural environment been considered in the allocation of land for agricultural purposes; and (b) what are the implications of the existing land allocation system and current land allocation pattern on the development and sustainability of agricultural land use. The study focuses on semi-arid land. The land allocation efficiency is determined by comparing the spatial heterogeneity of the land with the pattern of land allocation. The analysis is carried out at a sub-regional scale, and a local scale in twelve study areas. Changes over time are studied by comparing current land allocation patterns with those at Independence (1968). This study has identified two apparent weaknesses in the customary land management system. The frrst is in the capacity to ensure an efficient land resource distribution at a subregional level. The second is in the ability to ensure consistent land allocation practices at a local level. The study provides evidence that these shortcomings are now affecting the production environment and opportunities for development, and that changes in the tenure system are required. The study findings partly support a recent land po licy initiative proposing a gradual devolution in land management responsibilities to local level management systems, but also raise two major concerns. First, the land policy initiative does not address the shortcomings in sub-regional land management. Second, the inconsistent land distribution found at a local level does not support the notion that devolution will necessarily lead to more sustainable levels of land use within communities. In the wider debate on the agrarian transformation in Africa, this study adds to the body of knowledge in identifying specific shortcomings of indigenous management systems in land distribution, and their effects on sustainable agricultural development and land management. The study thus extends the more critical strand of thought on the role of local and indigenous land management systems in this process, and thus on the effectiveness of the devolution of resource management to community levels. The study also demonstrates that land sufficiency and quality are important issues in the process of sustainable intensification in small-scale land use systems, and question the wider applicability of the optimistic development model, which is primarily based on economic considerations. Lastly, the fmdings support the critical view on the applicability of the evolutionary theory of land rights in conditions similar to those in Swaziland. The fmdings of this study confIrm the importance of considering spatial scale and diversity in land use related studies, and show that any inference from one level of scale to another can be highly misleading.