A primary goal of behavioral interventions is to reduce dangerous or inappropriate behavior and to generalize treatment effects across various settings. However, there is a lack of research evaluating generalization of treatment effects while individuals with functionally equivalent problem behavior interact with each other. For the current study, the severe problem behavior of two participants with developmental disabilities was targeted for assessment and treatment. Results of a functional analysis indicated that for both participants, problem behaviors occurred when destructive behavior produced reinforcement of participant mands. The use of a multiple schedule that alternated between differential reinforcement of other behaviors and differential reinforcement of alternative behavior (mands) was sufficient to produce a significant reduction in problem behavior for one participant; the addition of differential reinforcement of alternative behaviors via compliance was necessary for the second participant. Extended treatment sessions, which focused on interaction between the two participants, involved one therapist concurrently implementing each participant's treatment while they were directed to play with one another (a situation that previously evoked problem behavior). Results show that each participant's treatment produced low to zero rates of problem behavior even during extended interactions.