Monthly first-contact data from the Groningen Psychiatric Case Register were used to study seasonal variation in the care-based incidence of psychiatric morbidity. Both overall and diagnosis-specific rates for a 15-year period (1976-1990) were examined. Regression analysis of overall rates revealed significant monthly deviations from the linear trend. Inspection of diagnosis-specific rates showed that the monthly number of first contacts varied most in patients with relatively mild psychiatric problems such as neuroses. Seasonal variation is believed to be a consequence of both fluctuations in true psychiatric morbidity and 'holiday effects' on supply of services and/or the inclination to ask for help. The relative influence of holiday effects is assumed to be inversely related to psychiatric severity.