Both direct and indirect evidence implicate excitatory amino acid neurotransmission in the aetiology of schizophrenia. The data are particularly suggestive for N-methyl-D-aspartate (NMDA) neurotransmission. Four of the six genes coding for subunits of the neural NMDA receptor have been mapped. We have studied segregation and allele sharing of markers in these four regions in a sample of southern African Bantu-speaking families multiply affected with DSM-III-R schizophrenia. This population was chosen because anthropological and linguistic data suggest that it has diverged from a small initial population within the past 1000 years, making shared genetic aetiology more likely. We find positive LOD score maxima of 0.876 at a marker D9S1838 on chromosome 9q34.3 near the NMDAR1 central subunit gene, 0.758 at marker D17S784 on chromosome 17q25 near the NMDAR2C potentiating subunit gene, and 0.453 at marker D12S77 near the NMDAR2B gene on chromosome 12p12 when analysing affected samples only. Only the region of NMDAR2A, on chromosome 16p13, can be excluded in this population. There is evidence of increased allele sharing on chromosomes 9p34.3 and 17q25 using APM. Multipoint allele-sharing analysis using GENEHUNTER does not reject possible effects on chromosome 9q34.3, but does not support any involvement of chromosome 17q25. We propose that the NMDA receptor may be involved in the genetic predisposition to schizophrenia in this population through covariation in several of the subunits, which is consistent with the genetic models of the inheritance of the disease.