Background. Although replication is the heart of science, psychiatric geneticists rarely have the opportunity to replicate findings, especially more than once.

Methods. This article reviews results from three independent family studies of schizophrenia on which one of us conducted diagnostic reviews: the Danish Adoption Study (DAS), the Iowa 500 non-500 family study (IFS), and the Roscommon Family Study (RFS). We utilized DSM-III or DSM-III-R criteria and meta-analysis techniques.

Results. The odds ratios (OR) in personally interviewed, first degree biological relatives of schizophrenic and matched control probands for schizophrenia, other non-affective psychoses (ONAP), schizotypal personality disorder (SPD), unipolar affective illness (UPAI), bipolar affective illness (BPAI), and anxiety disorders were homogeneous across studies. For alcoholism, ORs were significantly heterogeneous. Schizophrenia, SPD and ONAP strongly aggregated in relatives of schizophrenic probands with decreasing common OR estimates of 16·2, 5·0 and 4·0, respectively. The common OR for anxiety disorders was 1·1, indicating no familial co-aggregation. For UPAI and BPAI, the common ORs exceeded unity (1·3 and 1·9, respectively), although only the former was statistically significant.

Conclusions. Schizophrenia strongly aggregates in families and shares familial factors with SPD and ONAP but not anxiety disorders. The familial factors of aetiological importance for schizophrenia and affective illness may be weakly related. With the exception of alcoholism, the patterns of psychiatric disorders in relatives of schizophrenic and control probands in these three studies were sufficiently similar that, despite their methodological differences, they can probably be viewed as replications of one another.