This paper studies unethical behavior by groups and provides systematic evidence on how lying decisions are affected by group size and group gender composition. We conduct an online experiment with 1,677 participants (477 groups) where group members can communicate with each other via a novel video chat tool. Our key findings are that (i) larger groups lie more, (ii) all-male groups stand out in their proclivity to lie, (iii) already the first female in a group causes an honesty shift, and (iv) group behavior cannot be fully explained by members' individual honesty preferences.