One-locus models have been fitted to two sets of two-generational families, studied by Kallmann because each contained a schizophrenic twin proband; in one set each twin proband was in the nuclear group, in the other in the peripheral group. Allowing for a lognormal age of onset distribution and a dependence of ascertainment probability on age of onset, the following hypotheses were tested and rejected: ascertainment probability is a noncurvilinear function of age of onset on a log-log scale; the transmission of schizophrenia is via one Mendelian locus, and the transmission of schizophrenia is independent of parental type. The last hypothesis is more likely than one-locus Mendelian inheritance, but it too must be rejected.