Fire is a fundamental part of western US forest carbon (C) cycles.  We sought to characterize the effects of a large fire in the Siskiyou National Forest in Oregon, USA on the forms and flow of C in those soils.  After the fire, we sampled soils from 0-5 and 5-15 cm of burned and unburned plots, then separated the soils into light and heavy density fractions (LF, HF), followed by elemental and radiocarbon analysis of the fractions, soils, and charcoal.  There was a 15% net loss of soil C from the upper soil, which was not evenly distributed across the fractions: the LF sustained a 20% net C loss, while the HF-C showed a net gain of 5%.  There was a 20% net C gain in 5-15 cm, reflecting a 15% net C loss from the LF, and a 58% net increase in HF-C.  Radiocarbon values indicated that C loss from LF was not responsible for the net gain in HF-C at either depth.  There was a significant decrease (50 ‰) in the Δ¹⁴C of upper-soil HF from the burned plots, but no significant change in the 5-15 cm HF-¹⁴C.  Coarse charcoal Δ¹⁴C from the burned-plot soils indicated large inputs of wood >50 years old.  Mineral stabilization of fine charcoal from the surface could be responsible for C gains in the HF, and for observed ¹⁴C-depletion in the upper-soil burned-plot HF.  The Δ¹⁴C of 5-15 cm HF was similar to coarse charcoal, possibly explaining the lack of change in Δ¹⁴C even with large C gains.  Because charcoal is one of the more chemically recalcitrant forms of C in soil, and mineral stabilization tends to increase the mean residence time of soil C, it's likely that the C in these post-fire soils is substantially more stable than the pre-fire C distribution.