Verbal and written performance feedback for improving preschool and kindergarten teachers' treatment integrity of behavior plans was compared using a combined multiple-baseline and multiple-treatment design across teacher-student dyads with order counterbalanced as within-series conditions. Supplemental generalized least square regression analyses were included to evaluate significance. Maintenance of treatment integrity following termination of performance feedback was included and correspondence between treatment integrity and student behavior change was examined. Results suggested that both forms of feedback were effective for improving treatment integrity but that verbal performance feedback resulted in immediate and sustained improvements with moderate to strong correspondence with student behavior change.