The existing relationship between genetic markers of the cluster headache and the efficacy of lithium salts therapy was described in the present study. Thirty-five patients suffering from cluster headache, who were already typed for the HLA antigens, were studied. Typing was carried out with the microlymphocytotoxicity technique used by US National Institutes of Health. The patients were treated with lithium carbonate for a period of three months. Three parameters for evaluation of the efficacy of lithium therapy was used: the percentage of improvement, the wake-sleep rhythm and the pupil diameter measurement. The parameters were statistically (Student's t-test) evaluated and it was possible to separate two subgroups of patients: "responders" and "non-responders" to the lithium therapy. The phenotypical frequencies in the two subgroups was analysed using the chi 2 test, Data emerging showed a higher frequency of antigen HLA-B18 (23.8% versus 0%; p less than 0.005 pc less than 0.06) and of antigen HLA-A9 (42.9% versus 14.3%) in the "responders" subgroup. In the "non-responders" subgroup a higher frequency of antigen HLA-A1 (35.7% versus 14.3%) was found.