We examine voting by a board designed to mitigate conflicts of interest between privately informed insiders and owners. Our model demonstrates that, as argued by researchers and the business press, boards with a majority of trustworthy but uninformed “watchdogs” can implement institutionally preferred policies. Our laboratory experiments strongly support this conclusion. Our model also highlights the necessity of penalties on insiders when there is dissension among board members. However, penalties for dissent appeared to have little impact on the experimental outcomes.