Most information concerning possible cancer risks attributable to lifetime exposure to diagnostic x-rays comes from studies in which x-ray history was ascertained by interview or questionnaire, but little is known about the accuracy of such information. The authors assessed agreement between medical x-ray histories obtained through interview and by review of medical records from thyroid cancer case-control studies conducted in Sweden (1985-1992; 123 cases and 123 controls) and from members of a prepaid health plan in the United States (1986-1991; 50 cases and 50 controls). In both studies, substantial disagreement was found between the numbers of x-ray examinations reported in the interview and in the medical records. There was an indication of relatively poorer reporting among controls, particularly for certain types of x-ray examinations and for large numbers of such examinations. Estimates of the risk associated with exposure to diagnostic x-rays were similar, regardless of whether interview or medical record data were used, even though ordinal dose classifications based on the two sources differed considerably. In populations with a high frequency of exposure, spurious associations with numbers of x-ray examinations or estimated thyroid dose might arise because of differences in recall. However, in the present data, reporting errors by cases and controls seemed to be largely nondifferential.