To test the error catastrophe theory of aging we determined the error frequency of protein synthesis in several strains of cultured human fibroblasts at early and late passage. Error rates were calculated from analysis of native and substituted actins on two-dimensional gels of cellular proteins after induction of mistranslation by histidine starvation in the presence of histidinol. Early-passage cells from fetal, young, and old donors and cells from subjects with the Hutchinson-Gilford and Werner syndromes of accelerated aging had similar error frequencies. Late-passage cells from fetal, young, and old normal donors had similar or lower error frequencies than corresponding early-passage cells. No correlation was observed between error frequency, donor age, or maximal life span in vitro. We also examined an immortal cell line, simian virus 40-transformed W138 fibroblasts. These cells had a significantly elevated rate of mistranslation (2.8 +/- 0.2 x 10(-4))(+/- SEM) compared to their untransformed counterpart WI38 (0.6 +/- 0.1 X 10(-4)) or all diploid cells combined (1.1 +/- 0.1 x 10(-4)). Taken together, the data fail to support the error catastrophe theory of aging.