OBJECTIVE: To assess lineality in families of bipolar I probands, the authors used direct interviews of family members to reclassify families initially categorized as unilineal by family history. METHOD: The families of 1,800 treated bipolar I probands were screened by the family history method with multiple informants. If the proband had one or more affected sibs and one apparently unaffected parent, the parents (and then other available first- and second-degree relatives) were directly interviewed by psychiatrists. RESULTS: Of the 1,800 families screened, 56 were apparently suitable unilineal families with multiple affected members; 46 families were interviewed directly. After interviews with the parents, 12 families (26.1%) were found to be bilineal. Direct interviews of all available relatives in the 34 remaining families revealed that only 22 (47.8% of the 46 interviewed families) were unilineal or probably unilineal and 12 were probably bilineal. The probably bilineal families had a significantly higher proportion of siblings with unipolar disorder. In addition, the affected sibs from the probably bilineal families tended to have earlier onsets but had significantly fewer symptoms in the most severe depressive episode. CONCLUSIONS: Fewer than 50% of bipolar I families appearing unilineal according to family history were found to be unilineal by direct interviews. The phenotypic differences between the affected sibs from the probably bilineal families and those from the unilineal and probably unilineal families suggest differences in genetic mechanisms. These findings highlight the need to systematically assess lineality in all families considered for bipolar I linkage studies and support the preferential inclusion of unilineal families in linkage studies.Abstract Teaser