Young and elderly adults were required to generate or study synonyms or rhymes of stimulus words. In one experiment, multiple-trial free recall was used to assess possible developmental differences in the generation effect, a 48-hr delayed recognition task was used in the second experiment. There was a significant generation effect for young and old adults on both retention tasks. The effect, however, did not emerge for the elderly participants in free recall until the second or third trial. The implications of utilizing internal stimulus generation as a potent mnemonic device was discussed.