In 1994, the Kootenai River white sturgeon (Acipenser transmontanus) was listed as an Endangered Species as a direct result of two related observations. First, biologists observed that the white sturgeon population in the Kootenai River was declining. Second, they observed a decline in recruitment of juvenile sturgeon beginning in the 1950s with an almost total absence of recruitment since 1974, following the closure of Libby Dam in 1972. This second observation was attributed to changes in spawning and (or) rearing habitat resulting from alterations in the physical habitat, including flow regime, sediment-transport regime, and bed morphology of the river. The Kootenai River White Sturgeon Recovery Team was established to find and implement ways to improve spawning and rearing habitat used by white sturgeon. They identified the need to develop and apply a multidimensional flow model to certain reaches of the river to quantify physical habitat in a spatially distributed manner. The U.S. Geological Survey has addressed these needs by developing, calibrating, and validating a multidimensional flow model used to simulate streamflow and sediment mobility in the white sturgeon critical-habitat reach of the Kootenai River. This report describes the model and limitations, presents the results of a few simple simulations, and demonstrates how the model can be used to link physical characteristics of streamflow to biological or other habitat data. This study was conducted in cooperation with the Kootenai Tribe of Idaho along a 23-kilometer reach of the Kootenai River, including the white sturgeon spawning reach near Bonners Ferry, Idaho that is about 108 to 131 kilometers below Libby Dam. U.S. Geological Survey's MultiDimensional Surface-Water Modeling System was used to construct a flow model for the critical-habitat reach of the Kootenai River white sturgeon, between river kilometers 228.4 and 245.9. Given streamflow, bed roughness, and downstream water-surface elevation, the model computes the velocity field, water-surface elevations, and boundary shear stress throughout the modeled reach. The 17.5 kilometer model reach was subdivided into two segments on the basis of predominant grain size: a straight reach with a sand, gravel, and cobble substrate located between the upstream model boundary at river kilometer 245.9 and the upstream end of Ambush Rock at river kilometer 244.6, and a meandering reach with a predominately sand substrate located between upstream end of Ambush Rock and the downstream model boundary at river kilometer 228.4. Model cell size in the x and y (horizontal) dimensions is 5 meters by 5 meters along the computational grid centerline with 15 nodes in the z (vertical) dimension. The model was calibrated to historical streamflows evenly distributed between 141.6 and 2,548.9 cubic meters per second. The model was validated by comparing simulated velocities with velocities measured at 15 cross sections during steady streamflow. These 15 cross sections were each measured multiple (7-13) times to obtain velocities suitable for comparison to the model results. Comparison of modeled and measured velocities suggests that the model does a good job of reproducing flow patterns in the river, although some discrepancies were noted. The model was used to simulate water-surface elevation, depth, velocity, bed shear stress, and sediment mobility for Kootenai River streamflows of 170, 566, 1,130, 1,700, and 2,270 cubic meters per second (6,000, 20,000, 40,000, 60,000, and 80,000 cubic feet per second). The three lowest streamflow simulations represent a range of typical river conditions before and since the construction of Libby Dam, and the highest streamflow simulation (2,270 cubic meters per second) is approximately equal to the annual median peak streamflow prior to emplacement of Libby Dam in 1972. Streamflow greater than 566 cubic meters per second were incrementally increased by 570 cubic meters per second. For each