Traditionally female-dominated sectors are growing and male-dominated ones shrinking, yet sectorial male shares are not changing. Why? I embed a field experiment within the UK national recruitment program for social workers to analyse barriers to men's entry and the nature of men's sorting into female-dominated occupations. I modify the content of recruitment messages to potential applicants to exogenously vary two key drivers of selection: perceived gender shares and expectations of returns to ability. I find that perceived gender shares do not affect men's applications, while increasing expected returns to ability encourages men to apply and improves the average quality of the applicants. This allows the employer to hire more talented men, who consistently perform better on the job and are not more likely to leave vis-à-vis men with lower expected returns to ability. I conclude by showing that there is no trade-off between men's entry and women's exit among talented applicants, both at hiring and on-the-job, and thus the net impact of raising expected returns to ability for the employer is positive.