The occurrence and movement of water in the regional aquifer system that underlies the eastern Snake River Plain, Idaho, de- pend on the transmissivity and storage capacity of rocks that compose the geologic framework and on the distribution and amount of recharge and discharge of water within that frame- work. On a regional scale, most water moves horizontally through interflow zones in Quaternary basalt of the Snake River Group. In recharge and discharge areas, water also moves vertically along joints and interfingering edges of basalt flows. Aquifer thickness is largely unknown, but geophysical studies suggest that locally the Quaternary basalt may exceed several thousand feet. Along the margins of the plain, sand and gravel several hundred feet thick transmit large volumes of water. Regional ground-water movement is generally from northeast to southwest, from areas of recharge to areas of discharge. Re- charge is from seepage of surface water used for irrigation, stream and canal losses, underflow from tributary drainage ba- sins, and infiltration of precipitation. Aquifer discharge is largely spring flow to the Snake River and water pumped for irrigation. Major springs are near American Falls Reservoir and along the Snake River from Milner Dam to King Hill. Regional ground-water flow was simulated with numerical models. Initially, a two-dimensional steady-state model that in- cluded a nonlinear, least-squares regression technique was used to estimate aquifer properties. Later, a three-dimensional steady- state and transient model was used to replace the two-dimen- sional model. Three-dimensional model results indicated that average total transmissivity ranged from about 0.05 to 120 feet squared per second and vertical leakance ranged from about 3 x 10^-10 to 5 x 10^-6 feet per second per foot of aquifer thickness. The three-dimensional transient model was used to compare measured and estimated long-term changes in ground-water dis- charge and water levels with simulated values. Initial head con- ditions used in transient simulations were derived from a steady-state solution of estimated preirrigation hydrologic condi- tions. Transient simulations were 5-year stress periods beginning in 1891 and ending in 1980. Recharge for each stress period from 1926 to 1980 was estimated from surface-water irrigation, pre- cipitation, and streamflow records. Recharge for stress periods from 1891 to 1925 was based on the average value for stress peri- ods from 1926 to 1980 and was indexed to estimated irrigated acreages. Average annual tributary drainage-basin underflow for stress periods from 1891 to 1910 was calculated by using basin- yield equations. Underflow for stress periods from 1911 to 1980 was varied by use of streamflow records. Transient simulations reasonably approximated measured changes in aquifer head and ground-water discharge that re- sulted from use of surface water for irrigation. Irrigation with surface water peaked in about 1950; subsequent increases in irri- gation have been supplied largely by ground water. The three- dimensional model simulated water-level declines and reduced ground-water discharge caused in part by increases in ground- water pumping. The transient model was used to simulate aquifer changes from 1981 to 2010 in response to three hypothetical development alternatives: (1) Continuation of 1980 hydrologic conditions, (2) increased pumpage, and (3) increased recharge. Simulation of continued 1980 hydrologic conditions for 30 years indicated that head declines of 2 to 8 feet might be expected in the central part of the plain. The magnitude of simulated head declines was con- sistent with head declines measured during the 1980 water year. Larger declines were calculated along model boundaries, but these changes may have resulted from underestimation of tribu- tary drainage-basin underflow and inadequate aquifer definition. Simulation of increased ground-water pumpage (an additional 2,400 cubic feet per second) for 30 years indicated head declines of 10 to 50 feet in the central part of the plain. These relatively large head declines were accompanied by increased simulated river leakage of 50 percent and decreased spring discharge of 20 percent. The effect of increased recharge (800 cubic feet per sec- ond) for 30 years was a rise in simulated heads of 0 to 5 feet in the central part of the plain.