To evaluate the reliability of the quantitative extrapolation of the long-term effect of cancer therapies from somatic cells to germ cells, we compared the frequency of chromosome abnormalities in 303 lymphocytes from four individuals treated with radio- and/or chemotherapy 5-18 years earlier with the frequency in 422 spermatozoa from the same individuals. The mean frequency of structurally abnormal complements was much higher in germ cells than in somatic cells (P = 2.08 x 10(-6)). The fact that spermatogenic cells share cytoplasm is suggested as a possible factor in the increased viability of germ cells with chromosome aberrations. In addition, in spermatozoa the incidence of structural chromosome abnormalities was much higher in treated individuals than in controls (P less than 0.00060), while in lymphocytes no statistically significant differences could be observed. This observation and the apparent lack of relationship between individual frequencies in the two kinds of cells suggest that the long-term effect of anti-tumor treatments on germ cells cannot be extrapolated from the analysis of somatic cells.