Active wildfire seasons in the western U.S. warrant the evaluation of post‐fire forest management strategies. Ground‐based salvage logging is often used to recover economic loss of burned timber. In unburned forests, ground‐based logging often follows best management practices by leaving undisturbed areas near streams called stream buffers. However, the effectiveness of these buffers has not been tested in a post‐wildfire setting. This experiment tested buffer width effectiveness with a novel field‐simulated rill experiment using sediment‐laden runoff (25 g/L) released over 40 min at evenly timed flow rates (50, 100 and 150 L/min) to measure surface runoff travel length and sediment concentration under unburned and high and low soil burn severity conditions at 2‐, 10‐ and 22‐month post‐fire. High severity areas 2‐month post‐fire had rill lengths of up to 100 m. Rill length significantly decreased over time as vegetation regrowth provided ground cover. Sediment concentration and sediment dropout rate also varied significantly by soil burn severity. Sediment concentrations were 19 g/L for the highest flow 2‐month post‐fire and reduced to 6.9–14 g/L 10‐month post‐fire due to abundant vegetation recovery. The amount of sediment dropping out of the flow consistently increased over the study period with the low burn severity rate of 1.15 g L−1 m−1 approaching the unburned rate of 1.29 g L−1 m−1 by 2‐year post‐fire. These results suggest that an often‐used standard, 15 m buffer, was sufficient to contain surface runoff and reduce sediment concentration on unburned sites, however buffers on high burn severity sites need to be eight times greater (120 m) immediately after wildfire and four times greater (60 m) 1‐year post‐fire. Low burn severity areas 1‐year post‐fire may need to be only twice the width of an unburned buffer (30 m), and 2‐year post‐fire these could return to unburned widths.