Data gathered by the Educational Testing Service from 57,694 children in 264 schools were reanalyzed using the instructional group as the unit of analysis. Second-grade and sixth-grade children were tested in the fall and spring with the Metropolitan Achievement Test and either the Cooperative Primary Test or the Sequential Test of Educational Progress. Questionnaires were used to gather information about demographic characteristics, organization and implementation of programs, educational attitudes, pupil characteristics, and classroom activities. The data were analyzed using separate four-factor analyses of variance (sex by SES by instructional time by instructional emphasis) for each grade and program type. Results indicated that time spent in formal reading instruction was most strongly associated with gains in reading achievement. Instructional emphasis had less impact than instructional time, although in second-grade compensatory programs low instructional emphasis on skills combined with a maximum amount of time produced larger comprehension gains than did high emphasis on skills combined with maximum instructional time. Considered apart from instructional characteristics, sex and socioeconomic level did not influence gains in achievement. (AA)