Inbred mouse strains that lack the corpus callosum connecting the cerebral hemispheres in the adult differ from the C57BL/6J strain at several relevant but unknown loci. To identify at least one major locus that influences axon guidance, different strains showing phenotypically similar defects were crossed to test for allelism. If the F1 hybrid between two strains with the same brain defect is phenotypically normal, it is much more likely that the two strains will differ at fewer loci than will an acallosal strain and C57BL/6J. This approach proved to be very informative. Five reasonable models of inheritance involving two or three loci were assessed, and the data justified rejection of all but one hypothesis. A total of 479 mice were obtained from four inbred strains prone to absence of the corpus callosum (BALB/cWah1, BALB/cWah2, I/LnJ, and 129/ReJ), one normal strain (C57BL/6J), and 11 F1 hybrids among them. Because the size of forebrain axon bundles is generaly greater in mice with larger brains, and because whole brain size is certainly polygenic, the phenotypically normal groups were used to derive a standard index of the degree of corpus callosum deficiency relative to brain size. Results demonstrated clearly that the hybrid between BALB/cWah1 and 129/ReJ is normal, where as the crosses among the BALB/c substrains and I/LnJ yielded many mice with deficient corpus callosum. I/LnJ crossed with 129/ReJ also produced some animals with callosal defects. The data were consistent with a model in which the difference between BALB/c and 129/ReJ involves two loci, where as the defect in I/LnJ involves homo zygosity at three loci, which impairs development more severely. Furthermore, severe deficiency of the hippocampal commissure occurred in I/LnJ mice but never in crosses with other strains. Collisions of the fornix and anterior commissure occurred only in the BALB/c substrains and their F1 hybrid, and were independent of the CC defect.