Analyses of cytogenetic damage - chromosome aberrations (CA), micronuclei (MN), and sister-chromatid exchange (SCE) - are used to assess genotoxic exposure, on the supposition that higher levels of chromosome damage in peripheral lymphocytes reflect increased cancer risk. We attempt to test this hypothesis prospectively, by relating levels of cytogenetic damage to subsequent cancer morbidity in a cohort comprising 3190 subjects. All these subjects have been analyzed previously (1970-1988) for CA, MN, and/or SCE in studies of occupational and environmental exposure. The present paper describes the data base and assesses how the potential confounders smoking habits, sex, and age influence CA, MN, and SCE levels. Ten Nordic laboratories contributed data. In the present analyses, these data were treated separately to avoid the effects of interlaboratory differences. Point estimates from multiple regression analyses indicate that smoking may increase CA frequencies by up to 10-20% and SCE means by 5-8%, but that it has no effect on MN frequencies. Women had higher CA, MN, and SCE levels than men, but the sex effect was generally smaller than the effect of smoking. Age was positively associated with cytogenetic damage. Compared to the sex effect, the effect of a 10-year age increase was similar on CA, but less, 1-3%, on SCE. The amount of variation explained by the potential confounders taken together was generally low, often less than 20%. Thus, other still unknown factors must be the major sources of CA, MN, and SCE variability.