The Rialto?Colton Basin, in western San Bernardino County, California, was chosen for storage of imported water because of the good quality of native ground water, the known storage capacity for additional ground-water storage in the basin, and the availability of imported water. To supplement native ground-water resources and offset overdraft conditions in the basin during dry periods, artificial-recharge operations during wet periods in the Rialto?Colton Basin were begun in 1982 to store surplus imported water. Local water purveyors recognized that determining the movement and ultimate disposition of the artificially recharged imported water would require a better understanding of the ground-water flow system. In this study, a finite-difference model was used to simulate ground-water flow in the Rialto?Colton Basin to gain a better understanding of the ground-water flow system and to evaluate the hydraulic effects of artificial recharge of imported water. The ground-water basin was simulated as four horizontal layers representing the river- channel deposits and the upper, middle, and lower water-bearing units. Several flow barriers bordering and internal to the Rialto?Colton Basin influence the direction of ground-water flow. Ground water may flow relatively unrestricted in the shallow parts of the flow system; however, the faults generally become more restrictive at depth. A particle-tracking model was used to simulate advective transport of imported water within the ground-water flow system and to evaluate three artificial-recharge alternatives. The ground-water flow model was calibrated to transient conditions for 1945?96. Initial conditions for the transient-state simulation were established by using 1945 recharge and discharge rates, and assuming no change in storage in the basin. Average hydrologic conditions for 1945?96 were used for the predictive simulations (1997?2027). Ground-water-level measurements made during 1945 were used for comparison with the initial-conditions simulation to determine if there was a reasonable match, and thus reasonable starting heads, for the transient simulation. The comparison between simulated head and measured water levels indicates that, overall, the simulated heads match measured water levels well; the goodness-of-fit value is 0.99. The largest differences between simulated head and measured water level occurred between Barrier H and the Rialto?Colton Fault. Simulated heads near the Santa Ana River and Warm Creek, and simulated heads northwest of Barrier J, generally are within 30 feet of measured water levels and five are within 20 feet. Model-simulated heads were compared with measured long-term changes in hydrographs of composite water levels in selected wells, and with measured short-term changes in hydrographs of water levels in multiple-depth observation wells installed for this project. Simulated hydraulic heads generally matched measured water levels in wells northwest of Barrier J (in the northwestern part of the basin) and in the central part of the basin during 1945?96. In addition, the model adequately simulated water levels in the southeastern part of the basin near the Santa Ana River and Warm Creek and east of an unnamed fault that subparallels the San Jacinto Fault. Simulated heads and measured water levels in the central part of the basin generally are within 10 feet until about 1982?85 when differences become greater. In the northwestern part of the basin southeast of Barrier J, simulated heads were as much as 50 feet higher than measured water levels during 1945?82 but matched measured water levels well after 1982. In the compartment between Barrier H and the Rialto?Colton Fault, simulated heads match well during 1945?82 but are comparatively low during 1982?96. Near the Santa Ana River and Warm Creek, simulated heads generally rose above measured water levels except during 1965?72 when simulated heads compared well with measured water levels. Average