The manner in which trembling aspen (Populus tremuloides Michx.) forest duff moisture changes during the growing season was investigated in Elk Island National Park, Alberta, Canada. A calibration-validation procedure incorporating one calibration site with moisture sampling across three topographic positions was used to develop predictive models, which were subsequently compared with 12 validation sites across three vegetation types throughout the Park. Duff moisture was modelled against the Duff Moisture Code and Drought Code components of the Canadian Forest Fire Weather Index System. Spring, summer, and fall rates of duff moisture change differed (P< 0.050) during calibration, with moisture loss greatest in spring. Additionally, while moisture changes on the south-facing and crest topographic positions were similar during spring, moisture losses were greater (P< 0.050) at these locations compared with the north-facing landscape position. Correlation analysis indicated that duff inorganic content and bulk density were both related to duff moisture but were limited in importance compared with weather-based influences. When compared with predicted values obtained from calibrated models, moderate predictability of duff moisture was found (mean absolute error = 20.7%-54.2%). Relative to the national standard equations, unique but very different empirical relationships were developed between the Duff Moisture Code and Drought Code and the moisture content of the duff layer in aspen forest stands found in Elk Island National Park.