BACKGROUND: The single-item question of self-assessed health has consistently been reported to be associated with mortality, even after controlling for a wide range of health measurements and known risk factors for mortality. It has been suggested that this association is due to psychosocial factors which are both related to self-assessed health and to mortality. We tested this hypothesis. METHODS: The study was carried out in a subsample (n = 5667) of the GLOBE-population, a prospective cohort study conducted in the southeastern part of the Netherlands. Data on self-assessed health, sociodemographic variables, various aspects of health status, behavioural risk factors, and a number of psychosocial factors (social support, psychosocial stressors, personality traits, and coping styles) were collected by postal survey and structured interview in 1991, and mortality data were collected between 1991 and 1998. Cox proportional hazards analyses were used to calculate the association between self-assessed health and mortality, before and after controlling for the psychosocial variables. RESULTS: After controlling for sociodemographic variables, various aspects of health status, and behavioural risk factors, self-assessed health is still strongly associated with mortality in our dataset (Relative Risk [RR] of dying for 'poor' versus 'very good' self-assessed health = 3.98; 95% CI: 1.65-9.61). After controllin