Among 2, 195 reference citations, published during 1975 in ten major US medical journals, 634 (29%) were found to be erroneous on direct checking of the original source. The percentage of error within individual journals ranged from 14 to 50 per cent. Such a high error rate would seem to seriously diminish the usefulness of published reference lists and, possibly, raise questions about the accuracy of other portions of the literature also.