Human Ss were required to learn five simple paired associates. With speed of response as a measure and the presentation of the correct response item as reinforcement, both delay of reinforcement and activity during delay impaired training performance, conh n i n g a similar previous finding by Champion and Maride (1962). Interference was no greater for activity composed of common associations with the stimulus than for unrelated material. Statistical treatment of the parameters of fitted hyperbolic curves revealed that both delay and activity significantly reduced performance asymptote. The curve parameters also showed that activity was a significant determinant of growth rate although delay interval produced no significant growth effect. Lengthening the delay interval late in training lowered the performance of both active and inactive Ss;shortening the delay resulted in faster responding by only the active Ss.Learning experiments have shown that delay of reward is detrimental to the performance of subhuman Ss. Spence (1956) offered one attempt to explain this phenomenon by proposing that interfering responses develop during the delay period and become attached to the same stimulus as the response to be learned. He assumed that, numerically, the competing responses are a function of the length of the delay interval, so that the longer the delay the greater the interference and the poorer the performance. A study by Carlton (1954) supported Spence's hypothesis by showing that delay of reward was less detrimental for animals which were confined during the delay period in order to reduce the development of competing responses, than for unconfined animals.