Abstract: It is essential for the transformation of low-carbon agricultural practices and for achieving the national dual-carbon goals to study the spatial-temporal variation of agricultural carbon emissions and the factors influencing these emissions in Southwest China. We calculated agricultural carbon emissions, and explored the structure, intensity, and spatial-temporal variation of agricultural carbon emissions in the five provinces of Southwest China from 2005 to 2023. Moreover, the LMDI model was employed to identify the driving factors of agricultural carbon emissions across both the entire southwestern region and its individual provinces. The grey prediction model GM(1,1) was used to forecast emissions for the next seven years. The results indicate that during from 2005 to 2023, total agricultural carbon emissions in Southwest China ranged from 32. 1939 to 43. 1515 million tons, exhibiting an overall fluctuating upward trend, while the carbon emission intensity exhibited a V-shaped fluctuating decline. Throughout this period, the Tibet Autonomous Region consistently maintained Level V emission intensity, whereas the other four provinces reduced their intensity from Level V in 2005 to Level I in 2023. There were notable regional disparities, with Sichuan Province having the highest emissions, followed by Yunnan, Guizhou, Chongqing, and Tibet. Livestock farming constituted the predominant carbon source (48. 99% of annual average), followed by cropland soil utilization (26. 20%) and agricultural inputs (24. 80%). Agricultural production efficiency and labor scale had an inhibiting effect on emissions, whereas the level of agricultural economic development acted as a promoting factor; the impact of agricultural industrial structure varied contingent upon the rationality of provincial production systems. Projections suggest emissions will maintain growth during 2024—2030 with an annual average increase of 132, 900 tons relative to 2023 levels, with Sichuan demonstrating declining trends despite its high baseline emissions while other provinces exhibit gradual increases.