Gender imbalance in public sector hiring remains a persistent concern, yet research often overlooks how job advertisement features influence applicant self-selection. Thus, we focus on gender sorting in the public labor market, a mechanism potentially causing structural self-selection among job seekers. The study investigates how gender sorting affects applicant pools by examining gendered language and the gender of the contact person in job advertisements. We empirically test these mechanisms using a unique multi-source dataset consisting of actual job advertisements, a survey among recruiters issuing these job advertisements, and organization-level data (n = 1859). We obtain measures for gendered language using quantitative text analysis. Results from hierarchical linear models indicate that more feminine wording relates to a higher number and share of applications by women. Our research contributes to the literature, testing why women may apply less for some public sector jobs. The implications for research and policymakers and emphasize the relevance of gender sorting mechanisms in recruiting.