We used an analysis based on a geographic information system (GIS) to determine the amount of rearing habitat and stranding area for subyearling fall chinook salmon Oncorhynchus tshawytscha in the Hanford Reach of the Columbia River at steady-state flows ranging from 1,416 to 11,328 m3/s. High-resolution river channel bathymetry was used in conjunction with a two-dimensional hydrodynamic model to estimate water velocities, depths, and lateral slopes throughout our 33-km study area. To relate the probability of fish presence in nearshore habitats to measures of physical habitat, we developed a logistic regression model from point electrofishing data. We only considered variables that were compatible with a GIS and therefore excluded other variables known to be important to juvenile salmonids. Water velocity and lateral slope were the only two variables included in our final model. The amount of available rearing habitat generally decreased as flow increased, with the greatest decreases occurring between 1,416 and 4,814 m3/s. When river discharges were between 3,682 and 7,080 m3/s, flow fluctuations of 566 m3/s produced the smallest change in available rearing area (from -6.3% to +6.8% of the total). Stranding pool area was greatly reduced at steady-state flows exceeding 4,531 m3/s, but the highest net gain in stranding area was produced by 850 m3/s decreases in flow when river discharges were between 5,381 and 5,664 m3/s. Current measures to protect rearing fall chinook salmon include limiting flow fluctuations at Priest Rapids Dam to 850 m3/s when the dam is spilling water and when the weekly flows average less than 4,814 m3/s. We believe that limiting flow fluctuations at all discharges would further protect subyearling fall chinook salmon.