Sixteen subjects from the Naval Submarine Medical Research Laboratory participated in a dual-task study designed to measure processing requirements of a choice reaction time (RT) task. Two levels of choice RT stimulus-response (S-R)compatibility were tested with each of two tracking tasks to provide different levels of dual-task loading. In one tracking task, the target's temporal-spatial pattern was fixed; in the other, the target's path was a function of the subject's performance. In the choice RT task, compatibility was treated as a between-subjects factor, while the number of alternatives (set size) within a sequence was a within-subjects variable. Choice RT results indicated that compatibility and set size interacted; the increase in response latency as a function of set size was much greater when compatibility was low. An increase in choice RT response latency occurred when the secondary tracking task was added. Within a given compatibility level, this dual-task decrement was constant for all levels of set size; however, the magnitude of the dual-task decrement varied as a function of S-R compatibility, being greater when compatibility was low than when it was high. For these data, a model like Sternberg's (1969) stages model is seen to have more explanatory value than a pooled processing capacity model (e.g., Norman and Bobrow 1975).