Background: most studies on lifestyle changes in old age have been transverse. We have conducted a longitudinal study. Subjects: 112 non-institutionalized subjects were studied in 1984 and again 10 years later (ages in 1984 ranged from 53-82 years). Protocol: on each occasion subjects recorded in a diary their times of retiring and rising and of taking meals, during a 'typical week'. They also recorded whether they lived alone or with somebody. Analysis: the diaries were scored to establish any effects of age or living alone on the timing and variability of their lifestyle. Results: age was associated with changes in the sleep/wake schedule and mealtimes and a decrease of daily variation in these variables. When these changes were compared in subjects living alone and with somebody, the increase in time spent in bed and the decreases in variability of times of rising and meals were more marked in subjects living with somebody. Conclusions: a deteriorating body clock contributes to some of these changes, but an increasingly inflexible lifestyle will offset some of the effects of this decline in circadian rhythmicity.