Tree-age data in combination with fire scars improved inverse-distance-weighted spatial modelling of historical fire boundaries and intervals for the Darkwoods, British Columbia, Canada. Fire-scarred trees provided direct evidence of fire. The presence of fire-sensitive trees at sites with no fire scars indicated fire-free periods over their lifespan. Sensitivity analyses showed: (1) tree ages used in combination with fire-scar dates refined fire boundaries without biasing mean fire return intervals; and (2) compared with derived conservative, moderate and liberal thresholds (i.e. minimum burn likelihood cut-off values), fixed thresholds generated area burned estimates that were most consistent with estimates based on the proportion of plots that recorded historical fires. Unweighted and weighted spatial mean fire intervals (50–56 and 58–68 years respectively) exceeded dendrochronological plot-level (38-year) estimates based on fire scars only. Including tree-age data from fire-sensitive trees to calculate landscape-level fire interval metrics lengthened the mean return intervals, better representing historical high-severity fires. Supplementing fire scars with tree ages better reflects the spatiotemporal diversity of fire frequencies and severities inherent to mixed-severity fire regimes.