We show evidence that bipolar-disorder subtypes differ in their genetic architectures, suggesting that future subtype-specific analyses will reveal more-specific signals and biological mechanisms.