The paper argues that the distributive effects of technical change cannot be predicted except in the context of the institutional and social structure of the rural community, i.e. except in the context of a rural "system." Four kinds of considerations are seen to be fundamental to technology's ultimate impact on such systems: the characteristics of "abstract" technology (efficiency and factor intensity), the absolute magnitude and distribution of productive assets, the type and distribution of institutional services and local social customs and traditions. Technology, the distribution of assets and the distribution of institutional services are seen to determine the distribution of personal income. When coupled with traditional social customs, the distribution of personal income is transformed into the distribution of power. Three feedback loops complete the system: (1) the asset accumulation mechanism that links the distribution of income in time period t to the distribution of assets in period t + 1, (2) the distribution of power in period t that is assumed to influence the type and distribution of institutional services in period t + 1, and (3) the character of local customs and traditions that are assumed to be influenced by the characteristics of the technology and the distribution of assets with a lag of several periods. The analysis is made concrete by examining the effects of a particular innovation, the tubewell, on the distribution of income and power in rural communities of East and West Pakistan. It is concluded that the same technology, highly profitable and labor-using in both areas, will tend to strengthen the position of the dominant agricultural classes in Sahiwal District, West Pakistan and to weaken the position of the large farmer cum moneylender in Comilla, East Pakistan. The paper concludes with several generalizations about the characteristics of the various development situations that emerge when the weights of the four considerations in the system are varied. It is argued that the effectiveness of rural development policies and programs could be improved if they were based on more disaggregated typologies defined by different combinations of technology, asset distribution, institutional structure and traditional social organization.