We study whether board structure (board size, independence and gender diversity) in banks relates to performance. Using a broad panel of large US bank holding companies over the period 1997-2011, we find that both board size and independent directors decrease bank performance. Although gender diversity improves bank performance in the pre-Sarbanes-Oxley Act (SOX) period (1997-2002), the positive effect of gender diminishes in both the post-SOX (2003-2006) and the crisis periods (2007-2011). Finally, we show that board structure is particularly relevant for banks with low market power, if they are immune to the threat of external takeover and/or they are small. Our two-step system generalised method of moments estimation accounts for endogeneity concerns (simultaneity, reverse causality and unobserved heterogeneity). The findings are robust to a wide range of other sensitivity checks including alternative proxies for bank performance.