Background: Light-at-night has been shown in experimental studies to disrupt melatonin production but this has only partly been confirmed in studies of night shift workers. In this cross-sectional study, we examined the circadian variation of melatonin in relation to shift status, individual levels of light-at-night exposure, and diurnal preference, an attribute reflecting personal preference for activity in the morning or evening.Methods: One hundred and seventeen workers (75 night and 42 day) of both sexes, ages 22 to 64 years, were recruited from four companies. Participants collected urine samples from all voids over 24 hours and wore a data logger continuously recording their light exposure. Sociodemographic, occupational, lifestyle, and diurnal preference information were collected by interview. Concentrations of urinary 6-sulfatoxymelatonin (aMT6s), the main melatonin metabolite, were measured.Results: Mean aMT6s levels were lower in night [10.9 ng/mg creatinine/hour; 95% confidence interval (CI), 9.5-12.6] compared with day workers (15.4; 95% CI, 12.3-19.3). The lowest aMT6s levels were observed in night workers with morning preference (6.4; 95% CI, 3.0-13.6). Peak time of aMT6s production occurred 3 hours later in night (08:42 hour, 95% CI, 07:48-09:42) compared with day workers (05:36 hour, 95% CI, 05:06-06:12). Phase delay was stronger among subjects with higher light-at-night exposure and number of nights worked.Conclusions: Night shift workers had lower levels and a delay in peak time of aMT6s production over a 24-hour period. Differences were modified by diurnal preference and intensity of light-at-night exposure.Impact: Night shift work affects levels and timing of melatonin production and both parameters may relate to future cancer risk. Cancer Epidemiol Biomarkers Prev; 23(7); 1176-86. Ó2014 AACR.