Extensive and high-intensity forest fires can destroy surface vegetation, restart community succession processes, and are among the significant disturbances affecting the stability of forest ecosystems in central Yunnan. However, there is a lack of sufficient quantitative reports on the disturbance boundaries of these fire events and the subsequent vegetation recovery dynamics in the area. This study focuses on the burnt area from the fire that occurred on May 3, 1984, in the subtropical semi-humid evergreen broadleaf forest of Jizu Mountain, Dali. Based on Landsat multi-temporal images from 1986 to 2023, we attempt to quantitatively delineate the boundaries of fire severity for historical fire events using the difference Normalized Burn Ratio(dNBR). We also analyze and explain the spatial distribution pattern differences between areas of varying fire severity using topographic factors and multiple comparisons. The stages of vegetation recovery in the burnt area are quantitatively distinguished using remote sensing indices, including the Fractional Vegetation Cover (FVC), Normalized Burn Ratio (NBR), and Burn Recovery Ratio (BRR), in conjunction with time-series analysis and piecewise linear regression models. We examine time points in the vegetation recovery process by analyzing the temporal changes in climatic factors such as temperature and precipitation. Additionally, we assess the current degree of vegetation recovery degree in the burnt area using canopy height characteristics. The results indicate that the total burnt area in 1984 should be no less than 1423.71 hm². The burnt area is divided into four fire severity levels: low, moderate-low, moderate-high, and high, with area proportions of 63.39%, 30.73%, 5.85%, and 0.03%, respectively. Analysis reveals that areas with gentler slopes tend to have higher fire severity. The low and moderate-low fire severity areas are categorized into rapid recovery(1986 to 1996) and decelerated recovery stages (1996 to present), while the moderate-high and high fire severity areas are divided into rapid recovery(1986 to 1990), decelerated recovery(1990 to 1996), and stable(1996 to present) stages. However, climatic factors did not exhibit similar stage changes, and the temporal dynamics of remote sensing vegetation indices are more related to species iteration processes during vegetation succession. The average canopy height in all fire severity areas remains lower than that of the control area(26.4 m), indicating that the canopy structure and forest aboveground biomass are still undergoing recovery. After 40 years of recovery, the forest in the burnt area is still in the early to mid-stage of secondary succession, suggesting that the recovery of subtropical semi-humid evergreen broadleaf forests is a complex and long-term process.