Agricultural land conversion is critical for reconciling food security with carbon neutrality, yet the carbon emission impacts of specific conversion pathways across diverse regions remain insufficiently understood. This study provides a pathway-specific and regionally stratified assessment of carbon-emission changes from agricultural land conversion across China's nine agricultural regions from 1980 to 2020. By distinguishing paddy fields, dryland, and orchard land, we integrated land-use transfer analysis, carbon coefficient accounting, and Geodetector modelling to elucidate the mechanisms linking agricultural restructuring to carbon dynamics. The results show that China's agricultural land experienced limited net area change but substantial internal restructuring, highlighted by an orchard land decline-then-growth trajectory after 2000. Agricultural land conversion generated a net carbon-emissions increase of 3.67 & times; 108 t. Dryland-related conversion remained dominant, accounting for 54.62% of the four-stage average emission change and 61.85% of the cumulative change. Spatially, emissions exhibited a significant 'east-high, west-low' gradient, reflecting stronger carbonsource effects in eastern plains and coastal regions and weaker emissions or sink-enhancing effects in western regions. Driving mechanisms shifted after 2000: while topographic constraints remained fundamental, socioeconomic factors-specifically nighttime light intensity, population density, and GDP-became increasingly dominant in shaping emission differentiation. Sensitivity analysis confirmed that the total estimates, dryland dominance, and spatial pattern were stable under +/- 10% agricultural land coefficient scenarios. These findings transition land-use carbon research from aggregate accounting toward mechanism-oriented agricultural land governance, offering a scientific basis for region-specific low-carbon agricultural strategies under carbon neutrality goals.