Purpose: To establish the conditional use of requests for assistance as an alternative to escape maintained challenging behaviour. Method: The study involved a 12-year-old boy with a behaviour disorder. Experiment 1 involved the implementation of a reversal design comparing the learner's performance during 'easy' and 'difficult' addition problems. Experiment 2 involved the implementation of a quasi-experimental design to establish the conditional use of requests for assistance. Results: Conditional use of requests for assistance successfully competed with escape maintained challenging behaviour. Conclusions: Our results suggest that for a newly acquired response to be displayed conditionally, two events must occur. First, the individual must learn the skill that has been associated with requests for assistance. Secondly, the contingencies must be such that reinforcement value is greater for an independent response than for a response that involves a request for assistance.