For more than 20 years already, researchers from all over the world have tried to get insight into the genetic basis of psychiatric disorders such as schizophrenia (SZ) and bipolar (BP) disorder. Linkage and candidate gene association study results have led to a range of hypotheses about the pathogenesis of the disorders, but overall genetic findings have been inconsistent and not a single functional risk causing variant has yet been identified. Even genomewide association (GWA) studies in large samples, the most extensive and systematic interrogation of the genome thus far, seemingly have not brought the expected answers. A reasonable interpretation is that multiple rare variants, inherently linked with locus and allelic heterogeneity, confer a substantial proportion of susceptibility to the disorders. Also, structural variation might be an important factor and promising results are arising from copy-number variation (CNV) analyses. In this review we shortly touch on "old" results from linkage and association studies and critically review the design and "new" results of GWA and CNV studies. We discuss what can be learned from the past and how this knowledge can be used in future study designs.