Menstrual diary records were obtained from a total of 5257 women using nine different methods of contraception, one natural and eight hormonal. This paper presents a comparative analysis of their vaginal bleeding patterns. The analytic procedures follow the recommendations of a recent WHO workshop on bleeding pattern analysis, which involve dividing each subject's diary into successive 90-day reference periods, calculating ten indices for each period, and classifying women according to whether they have "clinically important" bleeding disturbances. In general, the findings of this analysis confirm those of previous studies. Women using the natural method, who were deliberately selected for the regularity of their menstrual cycles, averaged three bleeding/spotting episodes of length 5 days in each 90-day period, with very little variability within or between women. Subjects given a combined oral contraceptive had more regular patterns than any other treated group, with short (4-day) episodes and 23-24 day bleeding-free intervals. Progestogen-only pill users had more frequent, longer episodes and shorter, less predictable intervals than combined pill users. Contrary to widely-held beliefs, the progestogen-only pills produced fewer spotting days than the combined pills, and almost no spotting episodes at all. Nearly half of vaginal ring users experienced some menstrual disturbance in each period; their most common problems were irregular, infrequent or prolonged bleeding. Women using the long-acting injectable, depot medroxyprogesterone acetate, had totally unpredictable patterns, with infrequent but prolonged bleeding/spotting episodes. The incidence of amenorrhea rose from just under 10% in their first injection interval to over 40% in their fourth. The methods of analysis recommended by WHO in 1985 still require substantial refinement. Nevertheless, they are more sensitive than those used previously for WHO trials and produce an easily understood, clinically meaningful characterization of bleeding patterns.