Peripheral blood was irradiated with 2, 3, 4 or 5 Gy of X rays and was mixed with nonirradiated blood at five different dilutions to simulate partial-body irradiations. Analysis by FISH was performed using whole-chromosome painting probes for chromosomes 1, 4 and 11 in combination with a pancentromeric probe. Chromosome aberrations affecting the painted fraction were classified according to the PAINT nomenclature; other unstable aberrations affecting the unpainted material were also recorded. To evaluate the suitability of painting for dose assessment in partial-body irradiations, the ability of the u test and a proposed s test to detect the expected overdispersion and the similarity between the real doses and the doses estimated using Dolphin's approach were considered. For short-term biodosimetry, compared with solid-stained dicentric analyses, the suitability of FISH painting techniques for the detection of partial-body exposures is reduced, because of the decrease in the frequency of aberrations detected by FISH and in the number of cells with two or more aberrations. For reconstruction of past doses, when only complete apparently simple translocations in cells free of unstable aberrations were considered, the detection of the overdispersion and the accuracy of dose estimations were dramatically reduced. In a partial-body exposure, as the original dose increased, the whole-body dose estimated a long time after irradiation would tend to be lower, and the difference from the original dose would tend to be greater.