In this issue, Coryell and Zimmerman report no significant increase in the risk of schizophrenia or "schizophrenia spectrum" personality disorders in relatives of schizophrenics vs never-ill controls. This article attempts to evaluate critically and interpret their findings. Given the modest number of relatives studied, their negative findings regarding schizophrenia could easily be the result of sampling variation. Relatives were assessed using a general personality disorder instrument, and over two thirds were interviewed by telephone; this method may have a low sensitivity to detect key features of the schizophrenia spectrum such as inadequate rapport and suspiciousness. True population differences may, however, exist in the pattern of familial aggregation for schizophrenia and related disorders. The convincing demonstration of this would be of considerable interest and should lead to a constructive search for the source of the interpopulation differences.