N THEIR paper, "Effects of Errors in I Classification and Diagnosis in Various Types of Epidemiological Studies," Diamond and Lilienfeld' perpetuate a common error in the analysis of falsepositives and false-negatives. Since in so doing they appear to disprove a wellknown statistical result that misclassification tends to decrease true differences,2 a note in refutation is called for. In epidemiological studies,3 a patient who has a condition on examination (i.e., a positive) may report that he does not have it. He is called a falsenegative. Similarly, a patient who does not have the condition on examination (i.e., a negative) may report that he has it. He is called a false-positive. Notice that the population "at risk" of being false-negatives consists of persons who actually have the condition, and the population at risk of being false-positives are those who do not have it. Other definitions of false-positives and false-negatives lead to errors in interpretation, as will be shown later. The importance of these definitions is not in the interpretation of the original investigation, in which both the true status (on examination) and the reported status of each individual is known, but in the interpretation of other studies in which only the reported status is known. One of the most important situations in which the method is necessary is that in which some relatively inexpensive screening device (postal questionnaire, interview by nonmedical personnel, mass radiography) is used to estimate the prevalence of a condition whose true diagnosis is more expensive to establish.