Critics of ability grouping argue that, although the system permits instructional modification according to pupil ability level, it has harmful social and academic consequences for students in lower groups. This investigation asks two questions about the effects of grouping: First, does a student's within-class group rank affect his or her learning when individual ability and instructional content are held statistically constant? Second, do teachers utilize grouping in ways that have varied effects on student learning? OLS (ordinary least squares) regression test was performed on data collected through observation and testing in 12 Chicago-area first-grade classrooms. The analysis revealed that group rank has a small effect on student learning in the fall but that the effect dissipates as the year passes. Two subgroups of classes were extracted from the sample, and the same analysis was performed again. In "egalitarian" classrooms, where teachers appeared to de-emphasize the distinctions between groups, group rank had no effect on learning at any time during the year. In "elitist" classrooms, the distinctions between groups were highly pronounced, and even with instructional content controlled, group rank had a sizable impact on learning. That effect increased as the year went by so that by May, students in a top group learned 74 more words than students in a middle group, even though both were taught the same number of words. The study thus suggests that the consequences of grouping are not inherently detrimental but rather depend on how grouping is employed. Teachers might be trained to de-emphasize the stratification system within the classroom even as they use it to organize instruction appropriately. (Author/RH)