Bronchial micronuclei, small fragments of extra-nuclear DNA formed during cell division, provide a non-specific but quantifiable marker of DNA damage. Micronuclei have been used to assess carcinogen exposure and as an intermediate endpoint in chemoprevention trials. As part of an ongoing chemoprevention trial, heavy smokers underwent screening bronchoscopy, with biopsies taken at 6 standardized sites. Micronuclei counts were obtained for each site in each of the 40 volunteers found to have squamous metaplasia. Unlike squamous metaplasia, the average micronuclei counts among these heavy smokers were not associated with smoking history. Micronuclei counts were also not associated with the presence or extent of metaplasia. A striking degree of intra-individual variability was observed by comparing the micronuclei counts from different biopsy sites within individuals. The findings suggest that use of micronuclei from single sites may be misleading as a marker of carcinogen exposure or as an estimate of cancer risk. Serial measurements in individuals may provide the most useful information concerning carcinogenic exposure and the impact of chemopreventive agents.