A district-level reform model created by the Center for Data-Driven Reform in Education (CDDRE) provided consultation with district leaders on strategic use of data and selection of proven programs. Fifty-nine districts in seven states were randomly assigned to CDDRE or control conditions. A total of 397 elementary and 225 middle schools were followed over a period of up to 4 years. In a district-level hierarchical linear modeling (HLM) analysis controlling for pretests, few important differences on state tests were found 1 and 2 years after CDDRE services began. Positive effects were found on reading outcomes in elementary schools by Year 4. An exploratory analysis found that reading effects were larger for schools that selected reading programs with good evidence of effectiveness than for those that did not. (Contains 7 tables and 1 note.)