Previous family studies of good and poor prognosis schizophrenia are deficient in controlling for investigator bias and none have attempted to interview all available first degree relatives. This study attempted to correct for both of these deficiencies. One hundred fourty-four relatives of 25 poor prognosis and 28 good prognosis schizophrenics were systematically interviewed for psychiatric illness. Results showed a significant preponderance of affective disorder over schizophrenia in the families of patients with good prognosis and the reverse in the families of patients with poor prognosis. This finding is consistent with previous investigations. It confirms that family studies do not validate good prognosis schizophrenia as schizophrenia and suggests that most good prognosis cases are variants of affective disorder.