A method was designed to minimize the effects of measurement errors in studies of student growth. The Stanford Achievement Test Partial Battery, consisting of tests of paragraph meaning, word meaning, spelling, language, arithmetic reasoning, and arithmetic computation, were administered in a school system. Two eighth grade classrooms were chosen; one consisted of 25 middle ability students from a school using homogeneous grouping; the other included 22 pupils involving no grouping. Test reliability was estimated to be .96 and the standard error of measurement was approximately 3.0 months. To minimize the regression effect, the total of the four battery median scores was used as the basis for ordering the subjects within class groups according to achievement levels; averages of the battery medians were plotted for each third of the classes. After retesting one year later, the students appeared to be growing according to reasonable expectations. The growth lines of the top, middle, and low thirds of students became more widely separated as time passed. There was little support for the proposition that pupils of low average ability begin to level off before the end of elementary school. It was interesting that these students, although labelled as low achievers, grew at the same rate in three and one-half years as an average group of students.