The percentage of accidental pregnancies, the Pearl pregnancy rate and the life-table method have all been used to study the effectiveness of contraception in Belgium, using data from the Second National Fertility Survey (1971), which covered 3397 Belgian women in the age group 30-34 years. Though all three methods yield generally similar results, it is only by using the third method that we can obtain in an optimum way changes in contraceptive effectiveness by birth order and birth interval.Generally, contraceptive effectiveness is low, the cumulative failure rate for the first 12 months of all intervals amounting to 18%. This holds even for intrinsically extremely reliable methods, such as the pill (4%). The efficiency of contraception, however, improves with increasing duration of marriage. This is attributable not only to a shift towards more effective methods, but also to a better application of less reliable methods, such as withdrawal. It remains questionable whether the increase in effectiveness, especially of a method such as withdrawal, is not partially based on induced abortion.The findings also clearly reveal the dual aspect of the contraceptive behaviour in the sample: spacing until desired family size is achieved, and prevention thereafter.The efficiency of contraception decreases with increasing family size, whatever birth interval is considered.