At the pinnacles of organizations, comparative tests of unity of command and shared command are nearly impossible because only one individual sits atop most organizations. In organizations led by co-CEOs, however, such a test is possible because co-CEOs can truly share power. But do they? Our research pits the unity-of-command principle against the shared-command principle and finds overall support for the former, even within the co-CEO context. Our sample of 71 co-CEO pairs at publicly traded U.S. firms shows that increasing power gaps between co-CEOs are positively associated with firm performance. This positive association wanes and turns negative, however, as power gaps become very large. We conclude that whatever benefits the co-CEO structure might offer likely lie outside the shared command paradigm.