Using Bank of England voting data, we show empirically that members' votes are driven by heterogeneous individual assessments of the economy as well as their individual policy preferences. Estimates indicate that internal committee members form more precise assessments than externals and are also more hawkish. The estimates allow the first quantification of the gain due to information aggregation on monetary policy committees. The marginal gain from additional committee members tapers quickly after five members. There is no evidence of gains through externals' moderating internals' preferences. A relatively small committee of highly informed internal members emerges as a desirable committee structure.Keywords: Committees; monetary policy; private information.JEL Codes: E52, E58, D78 * We are very grateful to the editor Ricardo Reis and an anonymous referee for their insightful suggestions. We would also like to thank Tim Besley, Alan Blinder, Pablo Casas, Fabio Canova, Francesco Caselli, Antonio Ciccone, Greg Crawford, Thomas Cunningham, Wouter den Haan, Nathan Foley-Fisher, Francesco Giavazzi, Charles Goodhart, Matias Iaryczower, Kosuke Imai, Clare Leaver, Gilat Levy, Massimo Motta, Andrew Oswald, Morten Ravn, Karl Schlag, Kevin Sheedy, Jón Steinsson, Thijs van Rens, and members of Kosuke Imai's reading group for insightful comments on this and earlier versions of this work. We have also benefited from the suggestions of seminar participants at the