The upper middle Rio Grande Basin, as defined by the U.S. Army Corps of Engineers, extends from the headwaters of the Rio Grande in southwestern Colorado to Fort Quitman, Texas. Most of the basin has a semiarid climate typical of the southwestern United States. This climate drives a highly variable streamflow regime that contributes to the complexity of water management in the basin. Currently, rapid population growth in the basin has resulted in increasing demands on the hydrologic system. Water management decisions have become increasingly complex because of the broad range of interests and issues. For these reasons, the U.S. Geological Survey, in cooperation with the City of Albuquerque, New Mexico, conducted paired flow measurements at two cross sections to determine cross-sectional loss in the Albuquerque reach of the Rio Grande. This report statistically summarizes flow losses in the Albuquerque reach of the Rio Grande during the winter nonirrigation season from December 1996 to February 2000. The two previous flow-loss investigations are statistically summarized. Daily mean flow losses are calculated for the winter nonirrigation season using daily mean flows at three selected Rio Grande streamflow-gaging stations.For the winter nonirrigation season cross-sectional measurements (1996-2000), an average of 210 cubic feet per second was returned to the river between the measurement sites, of which 165 cubic feet per second was intercepted by riverside drains along the 21.9-mile reach from the Rio Grande near Bernalillo to the Rio Grande at Rio Bravo Bridge streamflow-gaging stations. Total cross-sectional losses in this reach averaged about 90 cubic feet per second. Regression equations were determined for estimating downstream total outflow from upstream total inflow for all three paired measurement studies. Regression equations relating the three daily mean flow recording stations also were determined. In each succeeding study, additional outside variables were controlled, which provided more accurate flow-loss measurements. Regression-equation losses between measurement cross sections ranged from 1.9 to 7.9 percent during the nonirrigation season and from about 5.9 to 6.4 percent during the irrigation season. Mean and median loss by reach length for all three daily mean flow stations and all three cross-sectional measurement reaches showed consistent flow loss per mile by season with allowance for nonideal river conditions for the initial measurement studies. Unsteady measurement conditions were reflected in the regression equation mean-square errors and ultimately in the change in daily mean discharge at the Rio Grande at Albuquerque gaging station during the measurement periods.