To the Editor.—Ìý Liebowitz et al,¹ in the July 1984 issue of the Archives, published preliminary results of a double-blind trial of phenelzine, imipramine, and placebo, based on a sample of 60 patients. They indicated that the study was still under way and that final results would be based on 120 patients. In 1978, the same group published similar partial analyses of a study of imipramine and behavior therapy in phobias,² later updated in the full sample.^3,4 I am writing to question whether this practice is a good research method and to encourage further discussion.There are two objections to violating the code of not analyzing and publishing clinical trials when they are partially completed. The first is that knowledge of the results might prejudice blindness and evaluation of the remaining cases in the study. This must be a risk in the present case, in which there