OBJECTIVE: Sleep-wake rhythm disturbances frequently occur in demented elderly and are of clinical relevance because they herald accelerated functional decline and institutionalization. Assessment of sleep-wake rhythm disorders is therefore of significant importance and can be performed by questionnaires or actigraphy, i.e., the recording of wrist activity. The present study investigates the relation of these two types of measurement by simultaneously assessing actigraphy and the Circadian Sleep Inventory for Normal and Pathological States (CSINAPS). METHODS: Seventy-eight elderly subjects, mean age 85+/-6 years, living in group care facilities of 12 homes for the elderly, wore an actigraph for two weeks. Caregivers completed the nurse informant CSINAPS. Spearman rank correlations and Mann-Whitney U tests were calculated over the equivalent sleep-wake rhythm parameters as derived from actigraphy and from the CSINAPS. RESULTS: Good correlations were found between questionnaire items about habitual timing of sleep and wakefulness and their actigraphic counterparts. Caregivers overestimated the actual sleep time between sleep onset and offset by 96 minutes. Questionnaire reports of sleep disturbances like wandering at night were also reflected in actigraphy parameters. However, the questionnaire and actigraphy variables correlate only modestly and may complement each other. In our study, both actigraphy and the CSINAPS seemed to miss the previously established high prevalence of sleep-disordered breathing (SDB) and leg movements during sleep (LM). CONCLUSION: The assessment of sleep and wake disturbances in demented elderly is best served by parallel use of a questionnaire like the CSINAPS and actigraphy. Moreover, if SDB and LM are a focus of interest, additional assessments are needed.