Experiments examining the issue of decay in short-term memory have assumed a single undifferentiated source of processing capacity which cannot be devoted to rehearsal when consumed in the processing of a nonverbal interpolated task. Three experiments reported here call this logic into question, since variations in difficulty in the nonverbal interpolated task failed to affect recall. Slight forgetting produced by a nonverbal interpolated task, relative to a no interpolated task control, was attributed to qualitative differences from performing two tasks simultaneously rather than only one. Results from the third experiment indicated that retrieval after a period of nonverbal interpolated activity is from primary rather than secondary memory.