Understanding past fire regimes requires reliable proxy data that record fire conditions and preserve them over time. The objective of this study was to determine if the oxygen isotope composition of charred bark samples (pyrogenic organic matter) could be used as proxy data to differentiate wildfires based on burn intensity. We collected charred and uncharred bark samples from three fire sites in northern Ontario, Canada that represented a gradient of fire intensity as depicted by Fire Weather Index (FWI) data. We hypothesized that the mean Δ¹⁸O_bark-char (the difference between δ¹⁸O of uncharred bark and a charred sample) would be greater for fires with higher intensities. Analysis of variance of Δ¹⁸O_bark-char indicated a significant effect of fire event (F = 73.6, p < 0.001), which explained 57.0% of the variance. A prescribed surface fire treatment (mean FWI = 9.5) had significantly lower Δ¹⁸O_bark-char than two natural crown fires (FWI = 21 and 27). These results demonstrate that Δ¹⁸O_bark-char differentiated moderate from high intensity fires in a similar manner to the FWI data.