Using a panel of 212 large U.S. bank holding companies over the period 1997-2004, we examine if board structure (board size, composition and gender diversity) in banks relate to their performance. After controlling for relevant sources of endogeneity (simultaneity, reverse causality and unobserved heterogeneity) via system generalized method of moments (GMM), we find that board structure in banks affect their performance. Particularly, the results show a negative association between board size in banks and their performance. We also find evidence of a negative relation between board independence and performance. In addition, our results support a positive association between gender diversity and bank performance. The study also shows that Sarbanes-Oxley Act of 2002 (SOX) has had an impact on the board structure and performance relation. Specifically, the negative effect of bank board size on performance is more pronounced in the post-SOX period. We also find that bank board structure is relevant particularly for banks with low market power and if exposed to external takeover threats.