SOME years ago, one of us (M. T. M.) wrote an article1 showing that the proband method of analysis, which was designed to test the theory that a trait was being inherited as a recessive character in human families, was incapable of giving the expected 25 per cent of recessive homozygotes unless the families with more than 1 affected child were counted as often as there were affected children in them, in which case the method became identical with Weinberg's so-called sibling method. The publication of several recent articles in the medical literature using the proband method has prompted us to bring up the matter again, and to show in what circumstances the proband method is capable of yielding the theoretic 25 per cent of affected children, or, rather, what corrections are necessary to make the method accurate. At the outset, we emphasize that we are not advocating the