Large-scale surface mining using valley fills has changed hydrologic storage and processes in the Ballard Fork Watershed in West Virginia. Total unit flow for the 2-year study period (November 15, 1999?November 14, 2001) on the Unnamed Tributary (extensively mined) (11,700 cubic feet per second per square mile) was almost twice that on Spring Branch (unmined) (6,260 cubic feet per second per square mile), and about 1.75 times that on Ballard Fork (downstream, partly mined) (6,690 cubic feet per second per square mile). Unit flow from the Unnamed Tributary exceeded that from the other two streams for all flows analyzed (5?95 percent duration). Unit flow from Ballard Fork exceeded unit flow from Spring Branch about 80 percent of the time, but was about the same for high flows (less than 20 percent duration). The proportional differences among sites were greatest at low flows. Spring Branch was dry for several days in October and November 2000 and for most of October 2001, and the Unnamed Tributary had flow throughout the study period. The increase in flows from mined parts of the Ballard Fork Watershed appears to result from decreases in evapotranspiration caused by removal of trees and soil during mining. During both years, evapotranspiration from the Spring Branch Watershed greatly exceeded that from the Unnamed Tributary Watershed during May through October, when leaves were open. Evapotranspiration from the Unnamed Tributary Watershed slightly exceeded that from the Spring Branch Watershed in February and March during both years. Evapotranspiration, as a percentage of total rainfall, decreased from the first to the second, drier, year from the Unnamed Tributary Watershed (from 61 percent to 49 percent) but changed little from the Spring Branch (from 77 to 76 percent) and Ballard Fork (73 to 76 percent) Watersheds. Precipitation and flow during the study period at three nearby long-term sites, the U.S. Geological Survey stream-gaging station East Fork Twelvepole Creek near Dunlow, West Virginia, and two National Oceanic Atmospheric Administration rain gages at Madison and Dunlow, West Virginia, were less than long-term annual averages. Relations observed among the three streams in the Ballard Fork Watershed during this study may not represent those in years when annual precipitation and flow are closer to long-term averages.