Per- and polyfluoroalkyl substances (PFAS) are widespread chemicals. Legacy PFAS have been phased out of production in most developed countries and emerging PFAS (short-chain PFAS and polyfluorinated compounds) are used as legacy PFAS alternatives. The effect of legacy and emerging PFAS on cytokine homeostasis in human remains poorly understood. This study aimed to evaluate the associations between legacy and emerging PFAS and cytokine profiles, and identify the main contributors to the disturbance of cytokine homeostasis. We quantified 21 PFAS in 198 Chinese women of childbearing age from 2015 to 2016. 13 cytokines were measured using the Meso Scale Discovery U-PLEX and V-PLEX platforms. The associations between PFAS exposure and cytokine levels were assessed using multiple linear regression (single-exposure), and Bayesian kernel machine regression (BKMR) models (PFAS mixture exposure). In single PFAS models, legacy and alternative PFAS were positively associated with Th1 and Treg cytokines, and negatively associated with Th2 and Th17 cytokines. For instance, each ln-unit increase in 6:2 chlorinated perfluoroalkyl ether sulfonic acid (6:2 Cl-PFESA), perfluorooctanoic acid (PFOA), and perfluorooctane sulfonate (PFOS) was associated with a decrease in IL-10 by − 0.228 (95% CI: − 0.336, − 0.120), − 0.153 (95% CI: − 0.277, − 0.030), and − 0.174 (95% CI: − 0.339, − 0.010), respectively. The BKMR model showed a significantly positive association of PFAS mixture with TGF-β and a negative association with IL-10. Overall, these results indicate that both legacy and emerging PFAS may affect the homeostasis of cytokines.