Information about the risks of diagnostic tests is incomplete, and its quality has not been studied. A set of criteria was devised and applied to 434 published reports to assess the quality of data about supposed adverse effects associated with 25 diagnostic procedures. Using these criteria, authors' conclusions concerning types of sequelae were supported by published findings in only 41% of reports. Information regarding frequency of these risks was upheld in 20%. Sound clinical judgment requires more dependable data than are presently available.