The clean energy transition has rendered end-of-life NdFeB magnets a strategic secondary resource for rare earth elements (REEs). However, existing recycling routes are either energy-intensive or generate hazardous effluents. This study pioneers a statistically optimized, airflow-controlled chlorination roasting–aqueous leaching (CRAL) process to selectively recover Nd, Pr, and Dy from magnet scrap, addressing critical gaps in reaction engineering and phase selectivity. Using a Box-Behnken Design within Response Surface Methodology (RSM), we systematically evaluated four interdependent parameters: time (60–180 min), temperature (200–400 °C), NH4Cl/NdFeB ratio (1–3), and airflow rate (0–4 LPM). Analysis of variance (ANOVA) revealed that temperature and airflow are the dominant control parameters, governing NH4Cl sublimation, local HCl availability, and the critical phase competition between water-soluble rare-earth chlorides (RECl3) and insoluble oxychlorides (REOCl). The quadratic models exhibited exceptional predictive capability (R2 > 0.996). Optimization predicted 327.5 °C, 2.25 h, a 2.075 ratio, and 3.45 LPM airflow. Experimental validation achieved 97.7% Nd and ~100% Pr recovery, while Dy recovery (24.2%) was limited by its lattice stabilization within the (Nd,Dy)2Fe14B phase. This work uniquely quantifies the previously understudied role of controlled airflow in enabling low-temperature (300 °C) selective REE extraction, producing an iron-rich stable oxide residue and minimizing aqueous waste. The statistically validated CRAL framework offers a robust, scalable, and sustainable pathway for circular REE supply chains.