Wildfire is a common natural disturbance agent in the Greater Khingan Range that significantly affects soil available phosphorus (AP) and soil microbial biomass phosphorus (MBP) in the burned area. This study aims to assess spatial and temporal changes of the AP and MBP in the burned area of Larix gmelinii. On this basis, the relationship between AP and MBP is further explored. Using the research method of "space-for-time substitution", we established a 28-year fire chronosequence with 80 sites over a small region of the Greater Khingan Range, selected three replicated plots in each burned area, and selected the nearby unburned area as the control plots. We measured the soil AP and MBP contents and topographic factors (altitude, slope position, slope aspect, slope gradient) for each plot. We analyzed the temporal and spatial evolution characteristics and the laws of AP and MBP in the burned area of Larix gmelinii. On this basis, the associated mechanisms of MBP and AP in the burned area of Larix gmelinii is further explored. After the fire disturbance, the contents of AP and MBP in the burned area initially decreased and then increased with the recovery time. In the initial stage of restoration, the MBP content of the burned area was significantly less than that of the unburned area, while the content of AP was significantly higher than that of the unburned area (P<0.05). The differences of AP and MBP contents in the burned area at different altitudes were significant (P<0.05), the differences of AP and MBP contents in the unburned area at different altitudes were not significant (P>0.05). The total explanatory power of the random forest regression model of soil MBP and AP in the burned area is about 84%, while that of the unburned area is about 60%, both models have reached an extremely significant level (P<0.001). In the random forest regression model of the unburned area, the slope position reached a significant level (P<0.05), the slope aspect, slope gradient, and altitude reached an extremely level (P<0.01). In the random forest regression model of the burned area, each explanatory variable reached an extremely level (P<0.01). The influence of fire disturbance on soil microorganisms is generally a long-term process and MBP is more sensitive than AP to fire disturbance. Fire disturbance will significantly strengthen the influence of terrain factors on AP and MBP, and change the relative importance of terrain factors on AP and MBP.