The authors, along with other investigators, postulate that viruses may be one of the causes of the syndrome of autism. Many diseases, especially those where a viral infection and autoimmunity is suspected, are being studied to determine whether an association with histocompatibility antigens (human leukocyte antigens--HLA) exists. The authors studied HLA in autism to see if a relationship exists. Twenty autistic children and their parents were HLA typed. The control group consisted of 575 potential donors for renal transplantation, 134 healthy subjects, and 48 persons of different families who married into one large family that had been HLA typed. The control subjects were from the same geographical area as the experimental subjects. Subjects were typed by a modification of the microlymphocytotoxicity tests of Terasaki and McCleland (1964). HLA-A2 was increased when compared to geographical controls, chi 2 = 5.020, p less than .05, and when compared to controls from the literature, chi 2 = 3.88, p less than .05. However, when chi 2 is corrected for the number of antigen specificities, significance is lost. No antigen was significantly increased in the mothers. HLA-A10 was significantly increased in the fathers, chi 2 = 5.947, p less than .02; however, significance did not remain after correction for the number of antigen specificities. These negative findings do not disprove an association because the numbers are so small. This small sample needs to be enlarged and replicated locally as well as in other geographical areas.