Spring and summer simulations were carried out using the Canadian Fire Behavior Prediction (FBP) and U.S. BEHAVE systems to study the role of vegetation and weather on fire behavior in the mixedwood boreal forest. Stands at Lake Duparquet (Quebec, Canada) were characterized as being deciduous, mixed-deciduous, mixed-coniferous, or coniferous, according to their conifer basal area percentage. Sampled fuel loads (litter, duff, woody debris, herbs, and shrubs) and local weather conditions (three different fire-risk classes) were used as inputs in the simulation. The predicted fire behavior variables were rate of spread (ROS), head fire intensity (RH), and area burned. Results from ANOVA testing showed that both weather and vegetation are not always significant, and the two prediction systems qualitatively attribute the explained variance to these factors differently. The FBP System selects the weather factor as the most important factor for all fire behavior variables, whereas BEHAVE selects the vegetation factor. However, three research burns located in Ontario revealed that BEHAVE was not well adapted to the mixedwood boreal region, whereas FBP predictions were quantitatively close to observed prescribed values. Extreme fire weather is confirmed as producing large and intense fires, but differences in fire behavior among stand types exist across the full range of fire weather. Implications of climate change, vegetation, and seasonal effects on fire behavior and the forest mosaic are discussed.©National Research Council of Canada. Abstract reproduced by permission.