Thirty-four healthy older adults (self-described "good sleepers") in their ninth decade of life (16m/18f, mean age 83.1) were compared to 30 young controls in their third decade (21m/9f, mean age 25.5) with regard to: (a) circadian and personality characteristics as measured by the Horne-Ostberg Morningness Questionnaire (HOM), Circadian Type Questionnaire (CTQ) and Eysenck Personality Inventory (EPI); (b) measures of habitual bedtime, waketime, and time in bed from a 2-week sleep diary; and (c) polysomnographic measures from a (post-adaptation) night of sleep recording in the laboratory. In almost all laboratory measures the older group slept poorly compared with the young, acquiring about one hour less total recorded sleep. The older group showed earlier habitual time of waking than the young, and showed higher (more "morning-type") scores on test instruments (HOM, CTQ-M) designed to assess morning-evening orientation. They also showed a lack of flexibility in sleep patterns (higher CTQ-Rs score) and less intersubject and intrasubject variability in habitual sleep timing compared to the young. Older subjects' morningness test scores were significantly associated with objectively measured sleep durations, with a tendency toward "morning-type" circadian orientation being associated with longer sleep.