Little is known about the course of chronic pain in the community. Such information is needed for the prevention and management of chronic pain. We undertook a 4-year follow-up study of 2184 individuals living in Grampian, UK to describe patterns and predictors of change in chronic pain over time. In October 2000, participants completed a postal questionnaire including case definition questions, the chronic pain grade questionnaire, the SF-36 and socio-demographic questions. Information from this questionnaire was compared to information collected from a similar questionnaire in 1996. A response rate of 83% was achieved for the follow-up study. The overall prevalence of chronic pain (pain or discomfort present either all the time or on and off for 3 months or longer) increased from 45.5% at baseline to 53.8% at follow-up. Seventy-nine percent of those with chronic pain at baseline still had it at follow-up. The average annual incidence was 8.3% and the average annual recovery rate was 5.4%. Individuals in the study samples who are in lowest quartile of SF-36 domains--physical functioning, social functioning and bodily pain at baseline--were more likely to develop chronic pain at follow-up, and respondents who were retired were less likely to develop chronic pain. Individuals in the study samples in the lowest quartile of SF-36 domains, bodily pain and general health at baseline, were less likely to recover from their chronic pain, as were those aged 45-74 compared with those aged 25-34. We concluded that chronic pain is a common, persistent problem in the community with relatively high incidence and low recovery rates. The lack of association between onset or recovery from chronic pain and most traditional socio-demographic factors, highlights the need to broaden the range of factors included in studies of chronic pain aetiology.