Summary Live fuel moisture content (LFMC) is a critical biophysical driver of wildfire risk, but it is currently unclear how dry do semi-tropical forests need to be in order to sustain fires. Here, we compared estimates from several satellites and products—sentinel-2, MOD09GA, and MCD43A4—and spectral indices to understand how does LFMC constrain fire activity in fire-prone Pinus yunnanensis forests from the subtropical mountain forests of Sichuan (SW China). We observed similar performance using the high spatial resolution sentinel-2 (R2 = 0.64–0.66) as well as with the coarser MODIS product MOD09GA (R2 = 0.62–0.77) using multiple linear regression. When hindcasting relationships between fire size and LFMC during 2001–2020, we observed that wildfire activity was critically associated with two thresholds of 127% and 115%. Our empirical approach for LFMC estimations can be immediately applied in mountain environments, which may greatly advance current fire danger estimates in these environments.