Bipolar disorder is a common, complex psychiatric disorder characterized by mania and depression. The disease aggregates in families, but despite much effort, it has been difficult to delineate the basic genetic model or identify specific genetic risk factors. Not only single gene Mendelian transmission and common variant hypotheses but also multivariate threshold models and oligogenic quasi-Mendelian modes of inheritance have dominated the discussion at times. Almost complete sequence information of the human genome and falling sequencing costs now offer the opportunity to test these models in families in which the disorder is transmitted over several generations. Exome-wide sequencing studies have revealed an astonishing number of rare and potentially damaging mutations in brain-expressed genes that could have contributed to the disease manifestation. However, the statistical analysis of these data has been challenging, because genetic risk factors displayed a high degree of dissimilarity across families. This scenario is not unique to bipolar disorder, but similar results have also been found in schizophrenia, a potentially related psychiatric disorder. Recently, our group has published data which supported an oligogenic genetic model of transmission in a family with bipolar disorder. In this family, three affected siblings shared rare, damaging mutations in multiple genes, which were linked to stress response pathways. These pathways are also the target for drugs frequently used to treat bipolar disorder. This article discusses these findings in the context of previously proclaimed disease models and suggests future research directions, including biological confirmation and phenotype stratification as an approach to disease heterogeneity.