Sleep duration is an important concept in epidemiological studies. It characterizes a night's sleep or a person's sleep pattern, and is associated with numerous health outcomes. In most large studies, sleep duration is assessed with questionnaires or sleep diaries. As an alternative, actigraphy may be used, as it objectively measures sleep parameters and is feasible in large studies. However, actigraphy and sleep diaries may not measure exactly the same phenomenon. Our study aims to determine disagreement between actigraphic and diary estimates of sleep duration, and to investigate possible determinants of this disagreement. This investigation was embedded in the population-based Rotterdam Study. The study population consisted of 969 community-dwelling participants aged 57-97 years. Participants wore an actigraph and kept a sleep diary for, on average, six consecutive nights. Both measures were used to determine total sleep time (TST). In 34% of the participants, the estimated TST in the sleep diaries deviated more than 1 h from actigraphically measured TST. The level of disagreement between diary and actigraphic measures decreased with subjective and actigraphic measures of sleep quality, and increased with male gender, poor cognitive function and functional disability. Actigraphically measured poor sleep was often accompanied by longer subjective estimates of TST, whereas subjectively poor sleepers tended to report shorter TST in their diaries than was measured with actigraphy. We recommend, whenever possible, to use multiple measures of sleep duration, to perform analyses with both, and to examine the consistency of the results over assessment methods.