Gender biases have been documented in many areas including hiring, promotion or performance evaluations. Many of these decisions are made by committees. We experimentally investigate whether committee deliberation contributes to gender biases. In our experiment participants perform a real effort task with subjective performance and rate the task performance of other participants. In a 3 $\times$ 2 design we vary the extent to which communication among raters is possible and whether or not the experiment is gender-blind. In the absence of communication there is gender bias. The bias disappears under minimal communication, but is strong and highly statistically significant under open communication. In the latter case 60 percent of ratings received by men are revised upwards after communication compared to only 25 percent of ratings received by women. As a consequence of the bias women are ranked on average three positions lower after deliberation. Additional treatments and analysis show that the incentives to agree typically present in committee deliberation work towards reducing the bias. By contrast, open communication seems to provide a platform for individuals with extreme views, who speak more often first in open deliberation than others. We then test two interventions for open deliberation. Randomizing the order of speaking does not reduce gender bias, but an information intervention where raters are informed of gender bias in prior sessions does.