In analyzing U.S. cancer mortality from 1933 to 1978, Doll and Peto speculated that recently recorded increases in mortality among those over age 64 were likely to be artifacts, reflecting deaths previously misattributed to such causes as nonspecific cancer, pneumonia, and senility and other ill-defined causes. We test this hypothesis by examining some age-specific, cause-specific mortality in persons 35 to 84 for the period of the 8-ICDA, 1968-1978, which corresponds to the last eleven years of the period studied by Doll and Peto. Our analysis reveals increasing trends in mortality from lung cancer, brain cancer and multiple myeloma in older whites and nonwhites along with increases in several potentially miscoded causes of death from 1968 to 1978. Thus, these increasing cancers in the elderly cannot be explained solely as artifacts. Further studies of trends in site-specific cancer mortality should include age groups through at least age 85. Continued monitoring of site-specific cancer mortality patterns, particularly among the elderly, remains crucial for developing preventive strategies to reduce cancers.