Two experiments investigated age differences in the encoding of associative information during a speeded naming task. In both experiments, semantically unrelated prime-target word pairs were presented 4 times, in either massed or spaced fashion, during the learning phase. An immediate or delayed test trial was presented following the fourth presentation. In Experiment 1, participants named both the primes and the targets. Younger and older adults showed similar benefits when naming targets that were part of a consistent prime-target pairing compared with targets presented with different primes at each presentation. In Experiment 2, participants named only the target word. Younger adults showed a benefit for consistently paired words, whereas older adults showed no benefit for consistently paired words. The results of the test trials showed a greater benefit for massed repeated words than for spaced repeated words at the immediate test and a reversed pattern at the delayed test. This spacing by test delay interaction was evident in response latency in Experiment 1 and in cued recall performance in Experiment 2.