Ash formed by the combustion of vegetation and the litter and duff layers may affect runoff and erosion rates in the period immediately following wildfires, but only a handful of studies have specifically measured its effect. Approximately 1 month after the 2005 Tarkio Fire in western Montana, we applied simulated rainfall for 1 h at a mean intensity of 7.5 cm h^–1 to six pairs of 0.5-m² plots in an area burned at high severity and with a continuous 1-cm to 3.5-cm ash layer on the surface. The ash was retained in one plot per pair and removed from the other plot with a fine brush. The ash layer provided an additional 1.5 cm of water storage capacity and protected the underlying mineral soil from surface sealing, so that the time of ponding in the ash-covered plots was 12 ± 10 min longer, the final runoff rate was 2.1 ± 0.4 cm h^–1 lower and the total infiltration was 2.0 ± 1.2 cm higher than in the plots without ash. The reduced runoff from the ash-covered plots in combination with reduced rainsplash detachment resulted in a 74% reduction in sediment yield relative to the plots without ash. Nine months after the fire, the ash was patchy and just a few millimetres thick, and a second set of simulations showed no significant differences in the time of ponding, final runoff rate, total infiltration or sediment yield between plots with and without ash. Ash may significantly reduce the runoff and erosion from severely burned areas in the period shortly after a fire. Consequently, the largest post-fire hydrogeomorphic hazard will occur after the ash layer has been removed by wind and water erosion, but before there has been substantial vegetative regrowth.