Nontransient 6th- and 8th-grade urban middle school students' achievement test scores were examined before (4th grade) and during (6th or 8th grade) enrollment in a performing ensemble. Ensemble participation (band, choir, none) and subject variables of socioeconomic status (SES) and home environment were considered. Fourth- and 6th-grade achievement tests consisted of Reading, Math, Citizenship and Science; 8th grade included Reading, Math, Social Studies, Science, and Language Arts. Analyses indicated significant differences yet small effect sizes for main effects of SES and ensemble participation. Higher SES students scored significantly higher on all subtests except 4th-, 6th-, and 8th-grade reading. Sixth-grade band students scored significantly higher than choir students and nonparticipants on every subtest of 6th- and 4th-grade achievement tests. Eighth-grade band students scored significantly higher than nonparticipants on 4th-grade Reading and Math and every subtest of the 8th-grade achievement test except Social Studies. Similar results for both cohorts suggest that band may attract higher achieving students from the outset and that test score differences remain stable over time. (Contains 2 figures, 3 tables, and 2 notes.)