A drip-type rainfall simulator was used to assess the effect of prescribed burning on infiltration rates and sediment production on rangelands near Kiboko, Kenya. Prescribed burns were applied in February and October 1982 and infiltration rate, sediment production and time to runoff were evaluated in June, September, October and November 1982. Infiltration rates were less and sediment production greater from the February burn which was applied prior to a long dry period as compared with unburned areas. Burned grasses resprouted but succumbed to drought before the end of the dry season. Infiltration rates were greater and sediment production lower from the October burn which was applied prior to the "short rainy season" as compared with unburned areas. Grasses in the October burned areas tended to have more leaves than stems whereas grasses in control areas had more stems than leaves.