This review concludes that textbook epidemiology has the epidemiologist applying "criteria of causality to the research before recommending clinical or public health actions" (Lilienfeld and Stolley, 1994), but public health studies (Fingerhut and Kleinman, 1990; Kellermann et al., 1993; McGinnis and Foege, 1993; Sloan et al., 1988; and Tardiff et al., 1994) have gone beyond any actual data analyzed to recommend a wide array of policies to be espoused either by physicians counseling patients or legislators confronting the problem of violence. Even if firearms were a cause of lethal violence, such recommendations should be no more automatic than identifying the HIV virus made clear how to treat AIDS or which public policies should be adopted to limit its spread. The recognition that nonfatal protective uses of guns must be known before a cost-benefit analysis of their value or utility could be made has not prevented publication of numerous public health studies that go well beyond modest research findings to recommendations for legislation that would restrict firearms access or even modifying firearm design in a variety of ways never tested by epidemiologists, regardless of the current feasibility of such testing. 68 references