Abstract: Forest fires occur frequently in China and have substantial impacts on ecosystem structure and functioning. Systematic identification of their long-term spatiotemporal evolution and regional differences is important for disaster prevention and mitigation and forest resource management. Based on forest fire statistics from 30 provincial-level administrative regions in mainland China, excluding Shanghai, during 2000-2023, this study applied the Mann-Kendall trend test and Pettitt change-point test to identify long-term trends and structural changes, and characterized interprovincial spatial differentiation using forest fire density, forest fire damage rate, forest fire grade structure, and stage-specific means. Remote-sensing cross-validation was conducted for eight representative provincial-level administrative regions using MCD64A1 forest burned-area data and FIRMS forest active-fire data within annual forest masks derived from the CLCD. Spearman rank correlation analysis was further used to examine the interprovincial associations between forest fire density and climatic and population factors. The results showed that forest fire activity in China transitioned from pronounced fluctuations at relatively high levels to sustained low levels during 2000-2023. Burned area and the number of forest fires entered statistically significant declining periods in 2012 and 2015, respectively, while 2011 was identified as a common statistically significant structural change point for both indicators. Compared with 2000-2011, the mean annual number of forest fires and burned area during 2012-2023 decreased by 74.5% and 89.1%, respectively. Forest fire occurrence and damage exhibited marked interprovincial differences, with relatively high fire density in some southern provinces and relatively high damage rates in several key northern forest regions before 2011. After 2011, the numbers of provinces with high fire density and high damage rates decreased markedly, although changes were not synchronous across provinces and indicators. Remote-sensing cross-validation showed that official forest fire counts and FIRMS forest active-fire counts exhibited consistent stage-wise changes in six of the eight regions, while officially reported burned area and MCD64A1 forest burned area showed consistent changes in seven of the eight regions. The pooled Spearman rank correlation coefficients were 0.515 and 0.604, respectively (p<0.001). At the national scale, annual mean temperature, annual precipitation, and population density were all significantly and positively correlated with forest fire density. Among the northern provinces, annual mean temperature and population density were significantly and positively correlated with forest fire density, whereas none of the three variables reached statistical significance among the southern provinces, indicating regional differences in the interprovincial associations between forest fire density and climatic and population factors. These findings provide a scientific basis for identifying forest fire prevention priorities, allocating fire-prevention resources, and implementing region-specific prevention and control.