Thousands of schools around the country have implemented whole-school reform programs to boost student performance. This paper uses quasi-experimental methods to estimate the impact of whole-school reform on students' reading performance in New York City, where various reform programs were adopted in dozens of troubled elementary schools in the mid-1990s. This paper complements studies based on random assignment by examining a broad-based reform effort and explicitly accounting for implementation quality. Two popular reform programs--the School Development Program and Success for All--were not found to significantly increase reading scores but might have been if they had been fully implemented. The More Effective Schools program was found to boost reading scores, but this effect seems to disappear when the program "trainers" leave the school.