One of the most persistent and urgent problems facing education policymakers is the provision of highly effective teachers in all of the nation's classrooms. Of all school-level factors related to student learning and achievement, the student's teacher is consistently the most important (Goldhaber 2002; Rockoff 2004; Rivkin, Hanushek, and Kain 2005). Even with substantial within-school variation in teacher effectiveness (Rivkin, Hanushek, and Kain 2005; Aaronson, Barrow, and Sander 2007), historically teacher evaluation systems have inadequately differentiated teachers who effectively improve student learning from lower-performing teachers. In Chicago from 2003 to 2006, for example, nearly all teachers (93 percent) received performance evaluation ratings of "Superior" or "Excellent" (based on a four-tiered rating system) while at the same time 66 percent of CPS schools failed to meet state proficiency standards under Illinois' accountability system (The New Teacher Project 2007). This study seeks to answer the following research questions about two waves of teacher evaluation reform in Chicago, a pilot (Excellence in Teaching Pilot or EITP) focused on rigorous classroom observations (2008-10) and a fully implemented evaluation system (REACH) that incorporates information from classroom observations and student assessment (2012-13 to present): (1) What does experimental evidence say about the effect teacher evaluation can have on school-level performance in mathematics and reading in elementary schools? and What does experimental evidence say about how teacher evaluation can differentially impact schools with different characteristics (for example, are there greater impacts in lower- or higher-achieving schools)? Findings from the first wave showed: (1) at the end of the first year of implementing EITP, schools improved student achievement in reading; and (2) more advantaged schools (i.e., schools that were high achieving prior to implementation, schools with lower rates of student poverty) tended to benefit the most from EITP. This finding suggests that an intervention such as teacher evaluation requires high levels of capacity in the school building in order to affect student learning. Analysis of the REACH data is still ongoing. Five tables and one figure are appended.