Major advances in understanding the physiology and genetics of circadian rhythm in the past decade challenge the researcher of sleep/wake disorders in Alzheimer's disease (AD) to distinguish patient characteristics stable across the course of illness ("traits") from characteristics that vary with stage of illness ("states"). A components-of-variance approach with a repeated measures model was used to examine the between-subjects variance over time ("trait") vs. within-subjects ("state") variance in 42 patients with probable AD followed, on average, over 2 years on actigraphic sleep/wake measures. Mental status scores indexed stage of illness. Actigraphic measures of sleep efficiency and circadian rhythmicity appeared predominantly "trait," with between-individual differences accounting for over 55% of variance compared to the less than 5% of variance related to stage of cognitive impairment. We discuss how "state-trait" analyses can be helpful in identifying areas of assessment most likely to be fruitful objectives of physiologic and genetic research on sleep/wake disturbance in AD.