We design an experiment to assess the effect of beliefs about gender in selecting oneself or a designated person to carry out a volunteering task. Participants in a volunteering task are given the option of selecting oneself or nominating someone from their group, and the group is described to them in terms of cartoons depicting women and men displaying different emotional states: happy, neutral or unhappy. We introduce a treatment consisting of gender priming, in which we elicit gender views with a set of 12 questions routinely used in social attitudes surveys to determine the degree of sexism of respondents. We find that women offer to volunteer more than men, and that while neither the emotional affect or the gender of the nominated person per se influence designation, men in the unprimed condition are more likely to choose the happy female face. Gender priming reduces designations and increases volunteering for all, but the treatment effect differs across genders: though both women and men are likelier to volunteer when primed, the men nominate fewer women across the spectrum of moods once gender primed, and the effect is stronger for the more sexist men, whilst women are reducing their delegation more uniformly once gender-primed, never nominate the neutral woman, and nominate the happy woman more often the less sexist they are. Our results provide evidence of both stereotyping by men and self-stereotyping by women: men are happy to pick any woman for the volunteering, though they display a preference for the happy woman, whilst women are both more sensitive to the mood displayed and prefer to pick women who might be happy to do it the less sexist they are. When it comes to actually carrying out the volunteering task, we find that, conditional on volunteering, women are more likely to actually follow through than men.