Objective: To analyse the prevalence of phantom (limb) pain over time and to analyse factors associated with phantom (limb) pain in a prospective cohort of amputees. Design: A multicentre longitudinal study. Patients: One hundred and thirty-four patients scheduled for amputation were included. Methods: Patients filled in questionnaires before amputation, and postal questionnaires six months, 1½ years and 2½ years to a maximum of 3½ years after amputation. Preoperative assessment included patients' characteristics, date, side and level of, and reason for amputation. The follow-up questionnaires assessed the frequencies of the experienced phantom pain, prosthetic use and walking distance. The occurrence of phantom pain was defined as phantom pain a few times a day or more frequently. Results: Pre-and postoperative questionnaires were available filled in by 85 amputees (33 females and 52 males). The percentage of lower limb amputees with phantom pain was the highest at six months after amputation, and of upper limb amputees at 1½ years. In general, more women than men experienced phantom pain. One and a half years and 2½ years after amputation the highest percentages of the lower limb amputees used their prosthesis more than 4 hours a day (66%), after that time this percentage decreased to 60%. The results of the two-level logistic regression analysis to predict phantom pain show that phantom pain was less frequently present in men (odds ratio (OR) ¼ 0.12), in lower limb amputees (OR ¼ 0.14) and that it decreased in due course (OR ¼ 0.53 for 1 year). Conclusion: Protective factors for phantom pain are: being male, having a lower limb amputation and the time elapsed since amputation.