In the clinical evaluation of new drugs, many investigators employ a simple crossover design, where half of the patients first receive the active drug for a given period of time, followed by a switch to an inert placebo for a similar period of time; the other half of the patients begin the trial on placebo, and are subsequently switched to the active drug. When the criterion measure consists of an over-all rating of, say, “Improved” or “Non-Improved”, the data lend themselves to a chi-square analysis. It is clear, however, that in many clinical circles, this analytic technique is being consistently misused, despite the fact that the various pitfalls and underlying assumptions have been widely emphasized (Lewis and Burke, 3). The purpose of this brief note is to demonstrate the proper use of chi-square, and also to show that, when incorrectly used, the clinical research investigator may increase the probability of prematurely discarding an effective drug.