We evaluated the potential effects of two different drawdown scenarios on rearing and migration habitat of subyearling fall Chinook salmon Oncorhynchus tshawytscha in John Day Reservoir on the Columbia River. We compared habitats at normal operating pool elevation with habitats at drawdown to spillway crest elevation and drawdown to the historical natural river elevation for two flows (4,417 and 8,495 m3/s). Using two-dimensional hydrodynamic modeling and a predictive habitat model, we determined the quantity and spatial distribution of rearing habitat and predicted water velocities. We predicted that the most habitat area would occur under normal pool elevation, but 93% of habitat was located in the upper third of the reservoir. Although less habitat area was predicted under drawdown to the spillway crest and the natural river, it was distributed more homogeneously throughout the study area. Habitat connectivity, patch size, and percent of suitable shoreline were greatest under drawdown to the natural river elevation. Mean cross-sectional water velocity and the variation in velocity increased with increasing level of reservoir drawdown. Water velocities under drawdown to the natural river were about twice as high as those under drawdown to spillway crest and five times higher than those under normal pool. The variability in water velocity, which may provide cues to fish migration, was highest under drawdown to the natural river and lowest under normal pool elevation. The extent to which different drawdown scenarios would be effective in John Day Reservoir depends in part on restoring normative riverine processes.