Records of the occurrence of vaginal bleeding were obtained from women using either a natural method of contraception or one of four types of hormonal contraceptive. The relationships between their bleeding patterns and a number of demographic variables were examined, with the aim of identifying subgroups of women who, if they used a particular hormonal method of contraception, would be likely to suffer more or less disruption to their bleeding pattern than the 'norm'. Within contraceptive method, bleeding patterns were more closely related to the women's geographical region of residence than to any other factor. Some of the differences between regions were consistent across contraceptive methods. European women tended to have more bleeding/spotting days than women in other regions; Latin American women had relatively short episodes and long bleeding-free intervals, whether they were using the ovulation method, combined pills or a vaginal ring. Other differences were method-specific. Women using combined pills in India or Pakistan had fewer spotting episodes than women using the same method elsewhere; those using progestogen-only pills had more. Regional variations in bleeding patterns were particularly marked among women using DMPA, and increased over time: by their fourth injection interval, 25% of European women had amenorrhea, as compared with 72% of subjects in North Africa. These findings need to be confirmed by carefully controlled studies of menstrual bleeding patterns and their acceptability in various ethnic groups. The results would be valuable in counselling new contraceptive acceptors, and could eventually guide the choice of methods for introduction into national family planning programmes.