The author of an article previously appearing in this journal attempted to evaluate the "practical efficacy" of different contraceptive methods, taking failures due to faulty use into account through a study of women requesting abortions. The work suffered from at least 2 major methodological errors: 1) the statistics on the contraceptive practices of women requesting abortions were compared with national statistics on contraceptive practices and not with statistics on the population at risk of requesting abortions at the author's medical practice. The fact that the sample of women requesting abortion was 1/3 foreign and 11% West Indian suggests doubt that their practices would be comparable to those of the national sample. 2) A less trivial error was that the authors combined figures for all women requesting abortions with all women accidentally becoming pregnant, tacitly assuming that the proportion of women translating an undesired pregnancy into a request for abortion is the same for users of all contraceptive methods. As every clinician knows, this is a false assumption: users of withdrawal, for example, are more likely to accept an unwanted pregnancy than are users of pills. The result would be overestimation of failure rates of more effective methods and underestimation of failure rates of less effective methods. The case illustrates the problem of presenting statistical and epidemiological work done by practitioners who lack the competencies required. Repeated exposure to erroneous use of statistics may tend to increase skepticism of other practitioners at any statistical evidence, even the most valid. The authors of the original article, responding to the criticisms, suggest that the findings of statistical studies are not always relevant to everyday practice, that the work was intended to provoke reflection on the unfavorable attitudes of practitioners to local contraceptive methods, and that another study on a similar topic will involve consultations with statisticians.