Pumped withdrawals compose most of the irrigation-water diversions from the Snake River between Upper Salmon Falls and Swan Falls Dams in southwestern Idaho. Pumps at 32 sites along the reach lift water as high as 745 feet to irrigate croplands on plateaus north and south of the river. The number of pump sites at which withdrawals are being continuously measured has been steadily decreasing, from 32 in 1990 to 7 in 1998. A cost-effective and accurate means of estimating annual irrigation-water withdrawals at pump sites that are no longer continuously measured was needed. Therefore, the U.S. Geological Survey began a study in 1998, as part of its Water-Use Program, to determine power-consumption coeffi- cients (PCCs) for each pump site so that withdrawals could be estimated by using electrical powerconsumption and total head data. PCC values for each pump site were determined by using withdrawal data that were measured by the U.S. Geological Survey during 1990–92 and 1994–95, energy data reported by Idaho Power Company during the same period, and total head data collected at each site during a field inventory in 1998. Individual average annual withdrawals for the 32 pump sites ranged from 1,120 to 44,480 acre-feet; average PCC values ranged from 103 to 1,248 kilowatthours per acre-foot. During the 1998 field season, power demand, total head, and withdrawal at 18 sites were measured to determine 1998 PCC values. Most of the 1998 PCC values were within 10 percent of the 5-year average, which demonstrates that withdrawals for a site that is no longer continuously measured can be calculated with reasonable accuracy by using the PCC value determined from this study and annual power-consumption data. K-factors, coefficients that describe the amount of energy necessary to lift water, were determined for each pump site by using values of PCC and total head and ranged from 1.11 to 1.89 kilowatthours per acre-foot per foot. Statistical methods were used to define the relations among PCC values and selected pumpsite characteristics. Multiple correlation analysis between average PCC values and total head, total horsepower, and total number of pumps revealed the strongest correlation was between average PCC and total head. Linear regression of these two variables resulted in a strong coefficient of determination R²=0 .9 86) and a representative K-factor of 1.463. Pump sites were subdivided into two groups on the basis of total head—0 to 300 feet and greater than 300 feet. Regression of average PCC values for eight pump sites with total head less than 300 feet produced a good correlation of determination (R²=0.870) and a representative K-factor of 1.682. The second group consisted of 10 pump sites with total head greater than 300 feet; regression produced a correlation of R²=0.939 and a representative K-factor of 1.405. Data on pump-site characteristics were successfully used to determine individual PCC and K-factor values. Statistical relations between pumpsite characteristics and PCC values were defined and used to determine regression equations that resulted in good coefficients of determination and representative K-factors. The individual PCC values will be used in the future to calculate irrigation- water withdrawals at sites that are no longer continuously measured. The representative K-factors and regression equations will be used to calculate irrigation-water withdrawals at sites that have not been previously measured and where total head and power consumption are known.