Data from a community-based four-year prospective study were used to investigate the relationship between marital disruption and long-term work disability. In 1990, all inhabitants of the municipality Ullensaker, Norway, belonging to six cohorts (20-22 years, 30-32 years, 40-42 years, 50-52 years, 60-62 years and 70-72 years) were sent a questionnaire. Of the 1,359 respondents who were working, married or cohabiting, and not older than 62 years, 1,115 (82%) returned a second questionnaire four years later (1994). Separation or divorce between 1990 and 1994 was related to long-term work disability in 1994 (OR = 3.02; 95% CI: 1.57-5.81), even after adjustments for age, sex, work characteristics, number of work hours per week, job satisfaction, body mass index, having pre-school children, smoking, physical leisure activity, emotional symptoms and musculoskeletal pain (all measured in 1990). Emotional problems evoked by the marital disruption may be part of the explanation.