Abstract: Using county-level data for Zhejiang Province (2000—2020), this study integrates hukou and resident population and outflow rates (82 counties) to depict spatio-temporal dynamics and identify drivers of rural population outflow. Spatial distribution and concentration are measured with Gini and GMI, and the evolution of outflow types is traced via a transition matrix. After Z-score standardization, we estimate OLS, spatial lag (SLM), and spatial error (SEM) models (OLS n=75; SLM/SEM n=73 under Queen contiguity with island units removed). Findings reveal a clear two-stage pattern-slow loss (2000—2009) followed by accelerated loss (2010—2020)-and a spatial configuration of 'hukou equalization-resident concentration-outflow shifting from concentrated to dispersed'. Conditional on covariates, per-capita fixed-asset investment is positively associated with outflow, while GDP per capita is negatively associated, evidencing a differentiated mechanism of 'incremental development promotes outflow, stock development retains population'. Rural employment share is significantly negative, and per-capita arable land is negative and marginally significant in spatial models, supporting a livelihood-anchoring effect; urbanization is positive and significant in spatial models, whereas vegetation cover is significantly negative. Model comparison shows SEM attains the lowest AIC/BIC and SLM a slightly higher log-likelihood; spatial parameters (ρ, λ) are not significant, indicating weak spatial interdependence rather than strong spillovers. We synthesize a testable mechanism chain-incremental pull, stock retention, livelihood anchoring, and weak spatial linkage-and propose an 'optimize mobility' policy mix: expand industry platforms and basic public services in core hubs, strengthen commuting links and industrial complementarity in persistently outflowing mountain-edge counties, and support talent return, remote work, and public-service baselines in transition counties.