Abstract: Theory linking labor inputs of irrigation agriculture to social organization is briefly reviewed. Labor input is distinguished into five tasks: construction, maintenance, allocation, conflict resolution, and organization of ritual. A sample of world communities is canvassed in search of structural variation. A rationale for studying these phenomena in a local, rather than a society-wide, context is presented. Types of ties of the locality with the larger system are explored. Several propositions about pervasive external linkages with local phenomena are presented. Millon’s results, showing no relationship between size of irrigation system and centralization, are challenged. It is found that often irrigation management roles are embedded in other socially powerful roles rather than forming part of a specialized bureaucracy. Conditions for role embeddedness are explored.