The high prevalence of chronic pain (duration >3 months) reported from different populations indicates a public health problem. Knowledge of the long-term course of chronic non-malignant pain is incomplete and scarce. This paper describes a follow-up of a cohort recruited from a survey in the general population. The cohort (n=214) consisted initially of individuals with widespread or located (neck-shoulder) pain or without chronic pain. The individuals were initially examined and replied to questionnaires on pain, social factors, lifestyle, medication and health care after two and 12 years. The deaths during the period were obtained from the population register. Complete data exist for 77% of the eligible individuals. After 12 years one-third of the individuals initially without pain reported chronic pain, and among those with initial chronic pain 85% still reported chronic pain. The number of painful areas was the strongest predictor of chronic pain 12 years later (OR 15.8; >3 locations vs. 0) whereas a social factor (having a close friend) decreased the risk (OR 0.44). The onset of chronic pain during the same period was related to the physical workload (work with bent positions; OR 5.31; yes vs. no). Mortality was significantly higher in the group initially reporting widespread pain compared with the other groups. The chronicity of widespread chronic pain supports early and intense intervention among individuals with located pain. The association between chronic widespread pain and increased mortality needs further investigation but may deepen the view of chronic pain as a public health problem.