The estimation of the magnitude of a dose of ionizing radiation to which an individual has been exposed (or of the plausibility of an alleged exposure) from chromosomal aberration frequencies determined in peripheral blood lymphocyte cultures is a well-established methodology, having first been employed over 25 years ago. The cytogenetics working group has reviewed the accumulated data and the possible applicability of the technique to the determination of radiation doses to which American veterans might have been exposed as participants in nuclear weapons tests in the continental U.S.A. or the Pacific Atolls during the late 1940s and 1950s or as members of the Occupation Forces entering Hiroshima or Nagasaki shortly after the nuclear detonations there. The working group believes that with prompt peripheral blood sampling, external doses to individuals of the order of about 10 rad (or less if the exposure was to high-LET radiation) can accurately be detected and measured. It also believes that exposures of populations to doses of the order of maximum permissible occupational exposures can also be detected (but only in populations; not in an individual). Large exposures of populations can also be detected even several decades after their exposure, but only in the case of populations, and of large doses (of the order of 100 to several hundred rad). The working group does not believe that cytogenetic measurements can detect internal doses from fallout radionuclides in individuals unless these are very large. The working group has approached the problem of detection of small doses (less than or equal to 10 or so rad) sampled decades after the exposure of individuals by using a Bayesian statistical approach. Only a preliminary evaluation of this approach was possible, but it is clear that it could provide a formal statement of the likelihood that any given observation of a particular number of chromosomal aberrations in a sample of any particular number of lymphocytes actually indicates an exposure to any given dose of radiation. It is also clear that aberration frequencies (and consequently doses) would have to be quite high before much confidence could be given to either exposure or dose estimation by this method, given the approximately 3 decades of elapsed time between the exposures and any future blood sampling.(ABSTRACT TRUNCATED AT 400 WORDS)