Malignant teratocarcinomas arise from developmentally totipotent normal stem cells. Whether the targets are embryonal somatic cells or germinal cells has long been a matter of controversy. Past experiments on teratocarcinoma induction by ectopic grafting of early rodent embryos or fetal germinal ridges have remained ambiguous because embryos ordinarily soon form germ cells, and parthenogenetic germ cells form "embryos." In order to interrupt the developmental cycle at its most telling point, day 6 (egg-cylinder stage) mouse embryos of genetically sterile types were grafted; in such grafts, only a terminal residue of totipotent embryonal somatic ("ectoderm") cells is available, and subsequent germ cell development is severely impaired. One graft series, from S1(J)/+ matings, comprised 25% S1(J)/S1(J) presumptive sterile embryos; these grafts formed tumors containing embryonal carcinoma cells as often (47%) as did control +/+ grafts (41%) on the same genetic background. In another series, from W/+ matings, tumors of the sterile W/W genotype were individually identified by means of a closely linked marker, phosphoglucomutase (PGM, EC 2.7.5.1; Pgm-1 locus), coding for electrophoretic enzyme variants and incorporated into the stock. Four tumors were obtained (out of 16) that had the PGM-1D phenotype diagnostic for W/W, and that also contained embryonal carcinoma cells. Therefore, the malignancy arises here in susceptible somatic embryonal stem cells at the terminal stage of their capacity for totipotency. Other teratocarcinomas-whether induced or spontaneous-of ostensible germ-cell origin by parthenogenesis may also depend upon development of the same somatic target cells before neoplastic conversion can occur. A general model based on these experiments is proposed for all malignancies: Malignant transformation of a particular kind of normal stem cell may be possible only when that stem cell has progressed to the threshold of further differentiation.