Water production from mountain watersheds depends on total precipitation input, the type and distribution of precipitation, the amount intercepted in tree canopies, and losses to evaporation, transpiration and groundwater. A systematic process was developed to estimate historic average annual runoff based on fire patterns, habitat cover types and precipitation patterns on the Tenderfoot Creek Experimental Forest. A fire history study in the Little Belt Mountains of central Montana indicates much of the experimental forest watershed burned in the 1700's and 1800's. Fire scars and existing timber stands on the 3,709 ha experimental forest show that two fires occurred in the 1700's and six in the 1800's covering more than 1,660 ha (45 percent) and 2,415 ha (65 percent), respectively. One small 32 ha stand on the experimental forest has not burned since 1580. The last major fire (206 ha) occurred in 1902 and three other small fires (covering only 19 ha) have been observed since the implementation of active fire suppression in the early 1900's. There has been no logging on this 3,709 ha forest of which 9 percent of the total area is composed of non-timbered meadows or rock outcrops. Annual water yield was estimated for Tenderfoot Creek Experimental Forest for the past 400+ years utilizing fire history, habitat cover types, current average annual precipitation and water yield/precipitation/cover type relationships. The maximum average annual runoff was estimated at 12,480 cubic dekameters (dams3) in the late 1500's based on 30 years of average annual precipitation (1961-1991). The 1581 to 1997 average water yield was estimated to be 11,680 dams3. The maximum water yield estimated for Tenderfoot Creek Experimental Forest, if all timber were removed, would be around 13,240 dams3. The minimum runoff if the entire forest was composed of mature lodgepole pine would be 11,230 dams3. The present yield of 11,360 dams3 is near the lowest yield of 11,250 dams3 estimated for 1873 and near the minimum possible for this experimental forest. During a wet year with all of the timber removed, runoff could be as high as 21,190, or in a dry year with most of the watershed covered with a mature forest as low as 5,620 dams3. On TCEF, fire suppression and succession appear to be creating conditions for a major fire event unless portions of the forest are removed by management actions that mimic historic vegetation patterns.