Background: SOC is associated with wellbeing and health. The Lundby Study is a cohort study of an unselected population (n=3563) in whom mental health and personality traits have been assessed since 1947, with follow ups in 1957, 1972, and 1997. Aim: To describe the relationship of Antonovsky’s 29-item sense of coherence scale (SOC) and its three subscales (comprehensibility, manageability, and meaningfulness) to mental health and mortality in an unselected middle-aged and elderly community cohort, controlling for gender, age, marital status, and socioeconomic status. Another aim was to analyse the three-factor structure of the SOC. Methods: Of the 1797 surviving subjects in 1997, 1559 participated in a semistructured diagnostic interview, and 1164 subjects completed the SOC questionnaire. Psychiatrists performed diagnostic evaluations. Collateral information was obtained from case notes and registers. Dates of death from 1997–2011 were obtained from the cause of death register. Results: SOC scores showed no sex differences, but were positively correlated with age. SOC scores were higher in married relative to unmarried participants and in blue-collar workers and self-employed individuals relative to white-collar workers. Total SOC and subscale scores were negatively correlated with depressive, anxiety, organic, and psychotic disorders. Male gender was positively correlated with comprehensibility and female gender was positively correlated with manageability and meaningfulness. Higher comprehensibility scores were correlated with lower mortality. Conclusions: SOC scores increased with age, were higher for blue-collar workers, and were lower for individuals with psychiatric disorders. Higher comprehensibility scores were associated with lower mortality. However, there was only weak evidence for a three-factor structure.