Previously reported linkage of bipolar affective disorder to DNA markers on chromosome 18 was reexamined in a large sample of German bipolar families. Twenty-three short tandem repeat markers were investigated in 57 families containing 103 individuals with bipolar I disorder (BPI), 26 with bipolar II disorder (BPII), nine with schizoaffective disorder of the bipolar type (SA/BP), and 38 individuals with recurrent unipolar depression (UPR). Evidence for linkage was tested with parametric and non-parametric methods under two definitions of the affected phenotype. Analysis of all 57 families revealed no robust evidence for linkage. Following previous reports we performed separate analyses after subdividing the families with respect to the sex of the transmitting parent. Fourteen families were classified as paternal and 12 families as maternal. In 31 families the parental lineage of transmission of the disease could not be determined ('either' families). Evidence for linkage was obtained for chromosomal region 18p11.2 in the paternal families and for 18q22-23 in the 'either' families. The findings on 18p11.2 and 18q22-23 support prior evidence for susceptibility loci in these regions. The parent-of-origin effect on 18p11.2 is confirmed in our sample. The delineation of characteristics of 'either' families requires further study.Although the etiology of bipolar affective disorder is unknown, strong support for an important genetic component comes from the results of family, twin, and adoption studies. 1 Linkage studies of bipolar disorder to date have provided suggestive evidence in favor of locus heterogeneity. Promising chromosomal regions suggested by recent linkage studies include regions on chromosome 18.Berrettini et al 2 first reported linkage of bipolar disorder to a region near the centromere on chromosome 18p in 22 families using the affected-sib-pair (ASP) method and the affected-pedigree-member (APM) method. Parametric LOD score analysis of all 22 families revealed negative LOD scores. However, individual families yielded LOD scores Ͼ1 assuming dominant or recessive genetic models. Confirmatory evidence for a bipolar susceptibility locus in this chromosomal region was found by Stine et al. 3 Both parametric LOD score analysis and ASP analysis supported linkage in their study of 28 families. In addition, the same study reported a second susceptibility locus on the long arm of chromosome 18 (18q21). Interestingly, linkage to loci on both 18p and 18q was strongest in those families, in which the father or one of the father's siblings was affected, suggesting a parent-of-origin effect operating in bipolar disorder. Gershon et al 4 re-analyzed the 18p marker data of Berrettini et al 2 by the sex of the transmitting parent. Although no kindred with limited paternal transmission was observed, ASP analysis yielded highly significant excess allele sharing in the pedigrees with mixed maternal-paternal transmission (in different pedigree branches) but not in pedigrees with exclusively maternal transmission confirmin...