Blood samples were collected from 192 exposed workers who participated in the cleanup after the April 26, 1986, nuclear reactor accident at Chernobyl, Ukraine. These samples, together with samples from 73 individuals living in Russia but not involved in Chernobyl cleanup activities, were collected during September 1991 to May 1996 and shipped to the U.S. for evaluation by three bioassays: cytogenetic analysis based on chromosome painting, HPRT mutation analysis and glycophorin A (GPA) variant analysis. Univariate statistical analyses of the results of each bioassay (including adjustments for age, smoking status and estimated precision of the bioassay) found greater frequencies of chromosome translocations and HPRT mutant T lymphocytes among the exposed individuals compared to the controls (P < or = 0.01). GPA analyses showed no significant difference for exposed compared to controls for either hemizygous, N/O, or homozygous, N/N, variant cell frequency. Multivariate analysis of variance of the subset of 44 exposed and 14 unexposed individuals with measurements from all three bioassays found elevated frequencies of chromosomal translocations and HPRT mutants, and reduced frequencies for both GPA end points among the exposed persons compared to the controls. However, none of these differences, considered singly or in combination, was statistically significant (although statistical power is low due to small sample sizes). Mean estimated dose, based on cytogenetic response, for those exposed was 9 cGy (range 0 to 51 cGy) and was less than that estimated by physical dosimetry (25 cGy). Correlation between the end points of the bioassays and estimated physical dosimetry was low (r < 0.2); the only significant correlation found was for physical dose estimate and dates worked at Chernobyl (r = 0.4, P < 0.01), with those working soon after the accident receiving greater estimated doses.