Forty-seven unrelated Danish patients considered to be manic-depressive, according to strict diagnostic, symptomatic, and course criteria, were typed for histocompatibility (HLA) antigens. Significantly more manic-depressive patients than controls were found to have HLA-A3, HLA-B7, and HLA-Bw16, while significantly fewer manic depressives than controls had HLA-B8. All eight of the patients with HLA-Bw16 were bipolar patients, and none were unipolar depressive patients. We emphasize the need to consider the results with caution in view of the large number of antigens considered and the relatively small number of patients involved. When statistical corrections are made for the large number of antigens investigated, only the difference between bipolar patients and controls remains significant. The best way to determine if our findings are really significant is to attempt to confirm them in other series of patients. The importance of utilizing strict symptomatic and course criteria for the selection and polarization of proband is stressed.