Arsenic (As) contamination in vegetables is a severe threat to human health. However, the evaluation of As relative bioavailability (As-RBA) or bioaccessibility in vegetables is still unexplored. The study sought to evaluate the As-RBA in commonly consumed ten leaf vegetables collected from As-polluted farmlands. Additionally, the As-RBA was determined using rat bioassay and compared with As bioaccessibility through five commonly used in vitro methods, including UBM (Unified BARGE Method), SBRC (Solubility Bioavailability Research Consortium), DIN (Deutsches Institut für Normung e.V.), IVG (In Vitro Gastrointestinal), and PBET (Physiologically Based Extraction Test). Results showed that the As-RBA values were 14.3–54.0% among different vegetables. Notably, significant in vivo–in vitro correlations (IVIVC) were observed between the As-RBA and the As bioaccessibility determined by the PBET assay (r² = 0.763–0.847). However, the other assays (r² = 0.417–0.788) showed a comparatively weaker relationship. The estimation of As-RBA using derived IVIVC to assess As exposure risk via vegetable consumption confirmed that As exposure risk based on As-RBA was lower than that the total As concentrations. Therefore, it was concluded that PBET could better predict the As-RBA in vegetables than other in vitro assays. Furthermore, As-RBA values should be considered for accurate health risk assessment of As in vegetables.