Comprehensive school reform efforts are an increasingly visible part of the educational landscape. Policymakers, educators, and researchers are eager to assess the effectiveness of these models, especially regarding their utility in the most troubled settings. This paper reports results from one such reform, the Talent Development Middle School (TDMS). Focusing on reading comprehension, the report analyzes data from two Philadelphia, Pennsylvania, middle schools that have been implementing TDMS and from two comparison schools. Hierarchical linear models suggest that TDMS has had overall positive effects on achievement. In general, otherwise similar students in Talent Development schools outperform comparison students, controlling for prior achievement. However, this overall TDMS effect is neither mediated nor accompanied by a positive effect of one recommended component of the TDMS approach -- peer-assisted learning. In addition, differential effects of Talent Development by grade were found. In the two Talent Development schools, which serve grades 5 through 8, some troubling trends have been found in sixth grade results. In light of this finding, implications for model development and future research are discussed. An appendix describes the TDMS model's core curriculum in detail. (Contains 7 tables, 2 figures, and 23 references.) (Author/SLD)