The author takes a critical look at the findings of the Katayama study of adverse reactions to ionic and nonionic contrast media. The author suggests that although the relative risks appear to be large and favor the use of nonionic media, serious adverse reactions were infrequent with both media types and risks were small on an absolute scale. In addition, the study demonstrates the potential for several serious information and selection biases. The accuracy of risk estimates are therefore called into question. Nevertheless, this study is extremely important in clarifying the risks associated with ionic and nonionic media during venography.