INTRODUCTION The efficacy of using sung words as a mnemonic device for verbal memory has been documented in persons with probable Alzheimer's dementia (AD), but it is not yet known whether this effect is related to music training. Given that music training can enhance cognitive functioning, we explored the effects of music training and modality (sung vs. spoken) on verbal memory in persons with and without AD. METHOD We used a mixed factorial design to compare learning (5 trials), delayed recall (30-min and, 24-hour), and recognition of sung versus spoken information in 22 healthy elderly (15 musicians), and 11 people with AD (5 musicians). RESULTS Musicians with AD showed better total learning (over 5 trials) of sung information than nonmusicians with AD. There were no significant differences in delayed recall and recognition accuracy (of either modality) between musicians with and without AD, suggesting that music training may facilitate memory function in AD. Analysis of individual performances showed that two of the five musicians with AD were able to recall some information on delayed recall, whereas the nonmusicians with AD recalled no information on delay. The only significant finding in regard to modality (sung vs. spoken) was that total learning was significantly worse for sung than spoken information for nonmusicians with AD. This may be due to the need to recode information presented in song into spoken recall, which may be more cognitively demanding for this group. CONCLUSIONS This is the first study to demonstrate that music training modulates memory of sung and spoken information in AD. The mechanism underlying these results is unclear, but may be due to music training, higher cognitive abilities, or both. Our findings highlight the need for further research into the potentially protective effect of music training on cognitive abilities in our aging population.