Old adults do not sleep as well as young adults. Mounting evidence demonstrates that sleep characteristics, such as sleep duration, structures and importantly the quantity and quality of specific sleep oscillations, decrease during aging and particularly strong in neurodegenerative disorders. These alterations might be related to functional consequences like memory impairment as observed in Alzheimer’s disease (AD). Some previous work indicated that sleep disruptions occur even before clinical onset of AD, therefore sleep disturbances were proposed to play a direct role in the progression of the pathology. However, early sleep changes and their functional consequences during the preliminary mild cognitive impairment (MCI) state, need to be investigated in greater depth. This study focused on physiological differences of sleep between healthy young, healthy old subjects and patients with mild cognitive impairment. Sleep architecture and more fine grained analyses on memory-relevant sleep parameter such as slow oscillations, sleep spindles and their functional interaction that is considered to be pivotal in memory consolidation, were examined during a 90-min nap. Furthermore, associations of nap-sleep characteristics with retention performance in a verbal and visuo-spatial memory task were evaluated. With regard to sleep physiology, significant differences between all three groups were observed for cross-frequency coupling measures that indicated higher fast spindle power during SO up-phases and stronger synchronization for young subjects relative to both older groups, and higher measures for healthy older subjects as compared to MCI patients. Makro-level sleep analyses revealed an age group difference for the duration of slow wave sleep only, with young subjects spending more time in slow wave sleep relative to both elderly groups. Similarly, analyses on sleep characteristics such as slow oscillation and fast spindle power indicated an age group difference: young subjects showed higher activity in these measures as compared to healthy old participants and MCI patients. Further, concerning pre to post sleep memory performance change, a group difference was only evident between young and both older groups in the verbal memory task. In conclusion, our evidence indicates that most differences in sleep physiology are observed in the course of (healthy) aging. Main decline with respect to sleep physiology that is observed in the early course of neurodegenerative pathology is only observed for the functional coupling between slow oscillations and sleep spindles, a mechanistic component considered crucial for the transfer of memories from hippocampus to cortical long-term storage networks.