OBJECTIVE: To examine the risk of multiple sclerosis in users of combined oral contraceptives. DESIGN: Cohort study conducted between 1968 and 1996 using diagnostic data supplied by general practitioners SETTING: General practices throughout the United Kingdom. POPULATION: Royal College of General Practitioners' Oral Contraception Study cohort of initially 46,000 women recruited during the late 1960s. METHODS: Directly standardised incidence rates of multiple sclerosis were calculated for current, former and never-users of oral contraceptives using first ever cases of multiple sclerosis reported by the general practitioners. The standardisation variables were age, parity, social class and smoking history. Five-year survival rates in the different contraceptive groups were calculated using standard life table techniques. RESULTS: One hundred and fourteen first ever cases of multiple sclerosis had been reported by November 1996 during 564,000 woman-years of observation. The incidence rate in both current and former users was not materially different to that in never-users. Although based on limited evidence there was no suggestion that the five-year survival was affected by a woman's use of combined oral contraceptives. CONCLUSIONS: These findings do not suggest a greatly elevated risk of multiple sclerosis during, or after, use of combined oral contraceptives.