The autosomal thymidine kinase (Tk) gene is frequently used as a target for the detection of mutations in mammalian cell cultures. In these cultures, one allele of the Tk gene contains an inactivating mutation (Tk-) so that single mutations in the remaining Tk+ allele cause functional loss of the Tk gene product. Except for their resistance to toxic thymidine analogues, Tk-/cells have no obvious in vitro phenotype, which permits the quantitative determination of Tk mutant frequencies in cell culture assays. Using homologous recombination in embryonic stem cells, we have created mice heterozygous for the Tk gene so that Tk gene mutations can be measured in vivo. However, the in vivo importance of the Tk gene remains unclear. Breeding Tk+/parent mice produced viable Tk-/-knockout (KO) offspring with reasonable efficiency. KO mice died within 7-8 months of birth. Among the several changes noted in KO mice, all mice that were observed histologically had kidney disorders (sclerosis of glomeruli), and many had vascular system disorders (polyarteritis). Also, when assayed at three months of age, the livers and kidneys of KO animals had a 7-8-fold increase in apoptosis without compensatory cell proliferation. There was no increase in spontaneous Hprt mutant frequency in spleen lymphocytes from KO mice. Our observations indicate that elimination of Tk gene function in vivo has an adverse effect on animal survival. This suggests that Tk-/-mutants generated by somatic mutation in Tk+/-mice may be under negative selection pressure. If this is the case, in vivo mutant frequencies for the metabolically active Tk gene may be underestimated.