In three experiments, we manipulated participants' perceived numerical status and compared the originality and creativity of arguments generated by members of numerical minorities and majorities. Independent judges, blind to experimental conditions, rated participants' written arguments. In Studies 1 and 2, we found that participants assigned to a numerical minority generated more original arguments when advocating their own position than did numerical majorities. In Study 3, an equal-factions control group was included in the design, and all participants were instructed to argue for a counter-attitudinal position. Those in the numerical minority generated more creative arguments than those in both the majority and equal-factions conditions, but not stronger arguments. We propose cognitive and social processes that may underlie our obtained effects and discuss implications for minority influence research. Copyright © 2008 John Wiley & Sons, Ltd.