Epidemiological studies have firmly established that cigarette smoking causes almost all cases of anaplastic and squamous cell bronchial carcinomas. It is equally well known, however, that many smokers do not develop lung cancer and, although the amount of tobacco smoked is undoubtedly, the dominant risk factor, it seems possible that other factors may also be involved. There is increasing evidence to support Boveri's old hypothesis that somatic mutation is an important event in the development of cancer and we have recently shown that cigarette smoke condensate (CSC) produces many dose-related lesions in the cellular DNA of cultured human lymphocytes as measured by sister chromatid exchange (SCE) induction. Cytogenetic studies had previously shown an increase in chromosomal aberrations in blood lymphocytes of heavy smokers relative to non-smokers, but little or no increase in SCEs. We have now measured basal and CSC-induced SCE rates in lymphocytes from different individuals and report that smokers have higher rates than non-smokers and that smokers with untreated lung cancer have consistently higher rates than their matched heavy smoking controls. These results are in keeping with the somatic mutational hypothesis and the epidemiological evidence, but also raise questions related to difficulties in smoking dosimetry and to possible variation among different individuals' responses to a common insult.