An experiment was designed to examine whether output interference determines long-term recency effects. Sixty one undergraduates memorized six lists consisting of six noun pairs. Each presentation of to-be-remembered pairs followed and preceded arithmetic tasks. After presentation of each list, 42 of the subjects were presented one member of a pair in the list as a probe and were asked to recall the other member of the pair. Following the probe recall, they were asked to free recall the entire list. The other 19 subjects engaged in ordinal free recall tasks. Long-term recency effects disappeared both in the probe recall and in the free recall following the probe recall. Under both conditions recall priority of recency portion also disappeared. Under ordinal free recall condition, long-term recency effects occurred when the recall started from recency portion. These results indicate that recall priority of recency portion is necessary for the buildup of long-term recency effects, suggesting that output interference determines the effects.