The availability and accuracy of data on cancer and smoking obtained from the next of kin of decedents was examined in connection with a cohort mortality study of steelworkers exposed to acid mist. Efforts were made to contact the next of kin of 372 decedents in the study, by one or two letters requesting return of a questionnaire and by telephone for nonresponders. The final response rate for next of kin was 59 percent, compared to 79 percent obtained with living cohort members in the same study. No living kin or no valid address for kin were found for 32.8 percent of the decedents. The response rate for next of kin was 40 percent for decedents who died before 1960, 47 percent for 1960 to 1969, 66 percent for 1970 to 1979, and 72 percent for deaths after 1980. Almost all next of kin could answer whether the decedent had ever smoked cigarettes or was a smoker at the time of death, but fewer respondents could answer more detailed questions concerning cigarette smoking. Medical data were sought from hospital records and doctors who signed death certificates. The cause of death was available for 67.8 percent of the decedents, but an adequate lifetime medical history was available for only 41.7 percent. Medical records provided information on smoking for only 24.2 percent. The next of kin correctly reported a history of cancer of 49 out of 59 decedents with a medical record of cancer. For 54 decedents for whom both next of kin and medical record data on smoking were available, 87 percent of the next of kin reported the same smoking status as in the medical records. The authors conclude that the difficulty in contacting next of kin and their ability to provide only limited information on smoking may make it difficult to perform accurate nested case/control studies.