The World Health Organization recommends the use of fixed reference periods for quantification of the incidence and severity of vaginal bleeding when patients use various forms of contraception. Ninety- and 110-day reference periods were used in the analysis of data from daily menstrual diaries kept by 72 healthy women in a one-year study of oral contraceptive agents containing ethinyl estradiol and either norethindrone or levonorgestrel. Analysis of bleeding patterns reported during both 90-day and 110-day periods revealed fewer days of bleeding and/or spotting overall with norethindrone than with levonorgestrel (e.g., a mean of 16.06 vs. 19.55 days, respectively, over the first 90-day period; P = .013) and significantly shorter bleeding and/or spotting episodes with the norethindrone preparation. This trend persisted when data were adjusted for a day-1 pill start. Using either method of analysis, duration of bleeding episodes was shorter among subjects taking norethindrone than levonorgestrel. Pills were missed in both study groups, but more women in the LNG/EE group missed from 1 to 3 pills in at least one cycle (31 vs. 21 in the NET/EE group). The between-group difference in bleeding events may be due to intrinsic hormonal differences in regimens or to the greater number of pills missed among levonorgestrel users.