Wildfire affects hillslope erosion through increased surface runoff and increased sediment availability, both of which contribute to large post-fire erosion events. Relations between soil detachment rate, soil depth, flow and root properties, and fire impacts are poorly understood and not represented explicitly in commonly used post-fire erosion models. Detachment rates were measured on intact soil cores using a modified tilting flume. The cores were mounted flush with the flume-bed and a measurement was made on the surface of the core. The core was extruded upward, cut off, and another measurement was repeated at a different depth below the original surface of the core. Intact cores were collected from one site burned by the 2010 Fourmile Canyon (FMC) fire in Colorado and from one site burned by the 2010 Pozo fire in California. Each site contained contrasting vegetation and soil types. Additional soil samples were collected alongside the intact cores and were analyzed in the laboratory for soil properties (organic matter, bulk density, particle-size distribution) and for root properties (root density and root-length density). Particle-size distribution and root properties were different between sites, but sites were similar in terms of bulk density and organic matter. Soil detachment rates had similar relations with non-uniform shear stress and non-uniform unit stream power. Detachment rates within single sampling units displayed a relatively weak and inconsistent relation to flow variables. When averaged across all clusters, the detachment rate displayed a linear relation to shear stress, but variability in soil properties meant that the shear stress accounted for only a small proportion of the overall variability in detachment rates (R² = 0.23; R² is the coefficient of determination). Detachment rate was related to root-length density in some clusters (R² values up to 0.91) and unrelated in others (R² values <0.1). The overall R² value improved and the range of exponents became narrower by applying a multivariate regression model where boundary shear stress and root-length density were included as explanatory variables. This suggests that an erodibility parameter which incorporates the effects of both flow and root properties on detachment could improve the representation of sediment availability after wildfire.