Men tend to self-select into leadership positions more often than women. This paper examines how confidence contributes to this phenomenon through two laboratory experiments. In Study 1,  we measure several dimensions of confidence and study their link with willingness to lead (WTL). For both men and women, we find that confidence is a stronger predictor of WTL when the leader’s role is advising followers, rather than taking decisions on their behalf. Building on this result, Study 2 identifies a causal effect of confidence on willingness to take on an advice-giving role, by manipulating confidence via randomized noisy performance feedback. We discuss the gender heterogeneity of this effect and its implications for policies targeting the representation of women in leadership.