To evaluate putative risk factors for Alzheimer's disease (AD), we conducted a case-control study with exposure assessment performed after disease onset. In this study, we evaluated the effect of recalled habitual napping according to its duration and examined whether APOE genotype modifies the effect. The subjects were 337 patients (144 men, 193 women, age at onset and the time of study (years): 69+/-10, 73+/-9) with a diagnosis of probable AD based on the NINCDS-ADRDA criteria. Two hundred and sixty spouses of the subjects (94 men, 166 women, age at the time of study: 69+/-9) served as controls. We asked retrospectively about habitual (3 or more days per week) napping and its duration observed between 5 and 10 years before the onset of AD for cases, and between 5 and 10 years before the time of the study for controls. The analysis revealed that limited napping for up to 60 min had an apparently protective effect against the development of AD, especially for carriers of the APOEepsilon 4 allele. By contrast, napping for more than 60 min increased the risk of AD morbidity among the carriers of the allele. Habitual napping may modulate or disturb the physiological functions of sleep and circadian time-keeping according to its duration, and this might be associated with some mechanism that leads to the development of AD.