A comprehensive statistical model is developed for the correct evaluation of contraception effectiveness. From the probability of a woman, using some contraceptive procedure, conceiving in a single cycle, the pregnant woman rate and the pregnancy rate are defined and calculated. This is used to infer the values of these rates from experimental trials, accounting for the number of followed women and for the period they are followed for, whatever kind of events may be counted up in the trials, either pregnant women or pregnancies. However, computing pregnancies can bias the results of a trial, since the conceiving women, supposedly with a greater risk of pregnancy, should be replaced in the sample by new ones, whereas computing pregnant women allows more objectivity, since those pregnant women can stay in the sample up to the end of the trial. Thus, a more realistic effectiveness rate can be deduced from the investigation.