The search for susceptibility genes for bipolar disorder (BD) depends on appropriate definitions of the phenotype. In this paper, we review data on diagnosis and clinical features of BD that could be used in genetic studies to better characterize patients or to define homogeneous subgroups. Clinical symptoms, long-term course, comorbid conditions, and response to prophylactic treatment may define groups associated with more or less specific loci. One such group is characterized by symptoms of psychosis and linkage to 13q and 22q. A second group includes mainly bipolar II patients with comorbid panic disorder, rapid mood switching, and evidence of chromosome 18 linkage. A third group comprises typical BD with an episodic course and favourable response to lithium prophylaxis. Reproducibility of cognitive deficits across studies raises the possibility of using cognitive profiles as endophenotypes of BD, with deficits in verbal explicit memory and executive function commonly reported. Brain imaging provides a more ambiguous data set consistent with heterogeneity of the illness. Keywords: bipolar disorder; genetics; endophenotype; neurocognitive function; brain imaging Bipolar disorder (BD) has a genetic basis with heritability of at least 80%. 1 A number of studies have attempted to map the susceptibility loci giving evidence of linkage in multiple genomic regions. In particular, findings in 4p, 4q, 8q, 10p, 12q24, 13q, 18p, 18q, 21q, and 22q have been supported by more than one study. [2][3][4][5][6][7][8][9][10][11][12][13][14][15][16][17][18][19] However, these studies have not identified any specific genes, and a recent metaanalysis of available genome scans showed no regions with a conclusive evidence of linkage. 20 The relatively slow progress of gene-mapping efforts is often explained by the 'complex' nature of the illness and of the genotype-phenotype relation. The complexity most likely includes heterogeneity, interactions of multiple loci, incomplete penetrance, as well as phenotypes resulting from environmental effects-either alone or in interaction with the genetic predisposition. The goal of psychiatric genetic research is identification of (heritable) variations in DNA sequence or function that influence behaviour or give rise to mental illness. By definition, in complex (multifactorial) traits, this link is not unique and specific. That is, a particular change in gene sequence does not necessarily lead to a specific behaviour, and similar behaviours (symptoms) may be due to distinct (sets of) genes. Moreover, we must consider the possibility that the path between the genotype and behaviour is not unidirectional. Emerging examples point not only to general effects of environment, for instance stress, on behaviour, but also to specific behaviours influencing gene function that can be transmitted onto the next generation. 21 No single method is considered clearly superior in identifying susceptibility genes for complex traits such as BD. Most researchers agree that a combination of strategies is most lik...