In allophenic (mosaic) mice produced from blastocysts injected with teratocarcinoma stem cells of the OTT 6050 transplant line, an unexpected coat phenotype led to the discovery that the tumor-lineage cells carried the steel gene (Sl(J)/+). Because steel also causes a macrocytic anemia, mosaics comprising both genetically anemic and normal (+/+) cells fortuitously provided a unique opportunity to examine in vivo the etiology of this anemia in light of previous results indicating that the lesion is extrinsic to the erythroid cells. The experiment differs from previous ones, which involved postnatal grafting, in that here hematopoietic stem cells of anemic and normal genotypes coexist throughout all developmental stages, confronted by tissues of the hematopoietic microenvironment that consist partly or solely of genetically normal cells. Therefore, the possibility exists that the anemia might be completely prevented rather than secondarily ameliorated. Moreover, variation in proportion of normal-strain cells in the hematopoietic supporting tissues could serve to "titrate" minimal requirements to promote normal erythropoiesis. Mice with mixed populations of steel- and normal-genotype cells in blood and other tissues were identified by means of independent markers specific for tumor vs. blastocyst strains of origin. The clinical blood picture of these mosaics proved to be indistinguishable from that of normal controls, even when only a small minority of cells in all tissues of one of the animals were genetically normal. Phenotypic blood normalcy was shown, by occurrence of the typical steel anemia among F(1) germ-line progeny of mosaics, not to be due to any change in the capacity of the mutant gene to elicit the anemia. The results from the mosaics thus demonstrate that the primary expression of the steel lesion is indeed in the hematopoietic microenvironment. However, they also reveal that a surprisingly small complement of normal cells there appears to be adequate to prevent this anemia permanently. The hypothesis is advanced that relatively short-range diffusible substances, produced by cells in the microenvironment and required for normal erythropoiesis, may account for the inductive effectiveness of small cell numbers.