In concurrent schedules of reinforcement relative response allocation or time allocation is often used as a measure of preference. In the present study response allocation to a switching key was developed as an alternative measure of preference for concurrently available food key stimuli. In experiment 1 the switching response was placed on a variable interval schedule of 30 seconds and proved to be a more sensitive measure of preference than food key responding. Experiment 2 investigated the effect of systematically varying the food key and switching key schedules; results were inconclusive, but switching performance remained the more sensitive measure of preference. In experiment 3 preferences for components of a multiple food key schedule were investigated but the switching key performance was at best only an ordinal indicator of preference. In experiment 4 switching key performance was more sensitive to reinforcement duration than was food key performance. The concluding experiment investigated switching performance as an indicator of preference for small frequent rewards versus large delayed rewards. The subjects' preferences for small frequent reward were modified by rewarding switching into large delayed food schedules and by punishing switching into small frequent food schedules. Results were discussed within the context of an operant analysis of self control.