Abstract The live fuel moisture content (LFMC) is an important precondition for wildfire activity, yet it remains challenging to predict LFMC due to the dynamic interplay between atmospheric and hydrological conditions that determine the plant’s access to, and loss of water. We monitored LFMC and a range of plant water-use traits (predawn and midday leaf water potentials [Ψleaf]), leaf traits (specific leaf area [SLA]), hydrological status (soil water content [SWC] in the shallow layer and full profile) and atmospheric variables (air temperature, vapour pressure deficit [VPD], CO2 concentrations) in a mature eucalypt woodland at the Eucalyptus Free-Air CO2 Enrichment (EucFACE) facility during a drought. We combined plant traits, hydrological status and atmospheric variables into a biophysical model to predict LFMC dynamics, and compared these with predictions of LFMC based on a satellite model and established relationships between Ψleaf and LFMC from pressure–volume curves. Predawn Ψleaf could be well predicted from changes in SWC, but variation in midday Ψleaf and LFMC were more responsive to atmospheric than hydrological variables. The biophysical model explained up to 89% of variability in LFMC and outperformed established approaches to predict LFMC. SLA was the single most important variable to predict LFMC, followed by VPD, which explained 33% of the remaining variability in LFMC. Our study demonstrates that the co-variation of plant traits and atmospheric and hydrological conditions affect LFMC during drought, suggesting a new way forward for predicting LFMC by combining biophysical and satellite-based models of LFMC with seasonal forecasts of meteorological and hydrological variables. Read the free Plain Language Summary for this article on the Journal blog.