The role of performance evaluation and testing in the development of an operational accountability system for urban public schools is discussed. Management science techniques and concepts and econometric methods are used to specify an accountability system, determine its data requirements, and evaluate alternative measures of systems performance. Empirical data, derived from surveys of administrative and teaching personnel in the Pittsburgh public school system and from the system's pupil and financial data files, are used in an examination of structural relationships among resource flows and the domains of administrative and teacher authority. Such analyses are shown to be prerequisite to the implementation of complete accountability systems in public education. (Author)