Functional analyses suggested that the disruptive behavior of three preschool children was maintained by escape from demands. While keeping the escape contingency intact, we conducted (a) a density analysis in which the children earned preferred items for task completion according to two schedules that varied in reinforcement density, and (b) a contingency analysis in which preferred items were delivered either contingent on task completion or provided noncontingently. Edible delivery (regardless of density or contingency) without escape extinction resulted in clinically significant decreases in the disruptive behavior of one child, and consistent but non-clinically significant reductions for a second child. An increase in compliance was observed for all three children. No effects of reinforcement density were detected at the parameters manipulated in this study, and no differences were observed between contingent and noncontingent reinforcement conditions. Escape extinction was eventually implemented with two children, and was effective with one of them. Additional interventions (instructional fading and embedding) were required to reduce the remaining child's rate of disruptive behavior. Our results suggest that noncontingent delivery of preferred items may be effective in decreasing escape-maintained behavior and promoting compliance for some children. (Contains 2 tables and 5 figures.)