The objectives of this study were to replicate age-related decrements in recollection and source memory, and to determine if repetition lag training improves recollection and whether these effects maintain and transfer to other tasks. Sixteen young adults and 46 healthy older adults participated, the latter of whom comprised hi-old (n = 16) and lo-old (n = 30) based on neuropsychological memory tests. All participants completed memory tests and questionnaires at baseline, and then half of the lo-old underwent nine days of repetition lag training while the other half engaged in a 9-day active control program. The memory tests and questionnaires were repeated immediately after the training or control program, and again three months later. The baseline data replicated well-established age-related decrements in recollection. Repetition lag training improved objective measures of recollection, eliminated the age-related recollection decrement, and these improvements maintained over three months. However, training did not transfer to any other objective test of memory thought to rely on recollection, or to any subjective memory measure. The results demonstrate for the first time that repetition lag training improves objective measures of recollection, eliminates recollection differences between younger and older adults, and that these gains maintain over a 3-month period posttraining. The lack of transfer to other tasks, however, indicates that training one type of recollection (for the studied modality in this case) does not affect other types of recollection (e.g., of an item’s recency). We suggest that recollection can be fractionated into many distinct types.