Teacher professional development (PD) is seen as critical to improving the quality of US schools (National Commission on Teaching and America's Future, 1997). PD is increasingly viewed as one of the primary levers for improving teaching quality and ultimately student achievement (Correnti, 2007). One factor that is driving interest in PD is rapidly mounting evidence that teachers vary greatly in their effectiveness (Nye et al., 2004). Another is a growing recognition that there is much for teachers to learn if they are to realize ambitious new content standards for students (Ball & Cohen, 1999; Borko, 2004). To this end, policymakers and funding agencies have directed considerable resources toward PD studies focused on improving teacher effectiveness (Birman et al. 2007). Despite recent shifts in research emphasizing the value of carefully designed experiments, the number of studies of teacher PD with rigorous designs and outcomes has lagged behind its student outcome counterparts. The purpose of the work reported in this proposal is to provide guidance to the research and policy community on the design of rigorous studies that have the power to draw strong causal inferences to establish the effectiveness of PD programs on valued outcomes. The authors' work considered a new generation of mathematics and reading knowledge outcomes designed to assess the types of content problems that teachers encounter in practice. Using these outcomes, they developed empirical estimates of design parameters such as the intraclass correlation coefficient (ICC) for teacher PD studies. Within this framework they first investigated the extent to which teachers' knowledge levels were clustered in schools and the extent to which this clustering varies by outcome domain and school context. Second, they investigated the extent to which uses of general design parameter estimates (e.g., ICCs irrespective of outcome) leads to indeterminacies in design. Third, they examined the extent to which the appropriateness of design parameters in mounting a well powered study were sensitive to the knowledge domain and the contextual features of studies. They then provide empirical estimates of domain- and context-specific design parameters to help researchers navigate methodological choices that lead to more effective PD study design. Four tables and one figure are appended.