In a genetic study of the first-degree relatives of 77 patients with delusional (paranoid) psychoses, the morbidity risks for schizophrenia, affective disorders, and atypical psychoses were evaluated using ICD-9 criteria. The prevalence of schizophrenia was 3.10 percent (4.12 percent with age correction to 40 years and 4.94 percent with age correction to age 60), which is higher than in investigations of paranoid psychoses, but lower than in studies of paranoid schizophrenia. The prevalence figure for affective disorders (age-corrected 3.04 percent for unipolar plus bipolar patients) is also intermediate to those for relatives of paranoid schizophrenics and paranoid psychotics. When the 77 index delusional patients were subdivided into axial syndromes (endogenomorphic-schizophrenic, endogenomorphic-cyclothymic, and organomorphic axial syndromes), two very homogeneous subgroups emerged. The endogenomorphic-schizophrenic subgroup showed high rates of schizophrenic secondary cases, whereas the endogenomorphic-cyclothymic subgroup showed high rates of affectively disordered secondary cases. The third organomorphic subgroup showed a high prevalence of atypical psychoses among first-degree relatives. Thirty-seven of the 77 index patients could not be assigned to any axial syndrome; their first-degree relatives had an increased prevalence of schizophrenia, but affective disorders were no more frequent than in the normal population. These data suggest that the heterogeneous group of paranoid psychoses can be meaningfully subdivided by use of axial syndromes which are viewed as representing "basic" disturbances underlying delusional symptomatology.