Restoration of riparian vegetation along large rivers is complicated by the patchiness of the habitat and by conflicts with the societal need to control flooding. The Sacramento River Project, led by The Nature Conservancy in northern California, is testing whether it is possible to restore native forest along a large river without removing flood control. We conducted a post-hoc analysis of monitoring data collected by the project on 1–4-year old plantings of 10 native trees and shrubs at five sites. Two questions of general interest were: Can one identify types of species or sites that are especially suitable for restoration in such riparian habitats? To what degree must sites be treated as mosaics of patches, with different types of patches that are suited to different species? Plant performance as measured by height was better in species of Salicaceae or in species planted as cuttings than in species of other families or in species planted as seedings or seeds. Three within-site factors, land form, soil depth to a buried layer of sand or gravel, and soil texture, affected the growth of several species, indicating that sites do need to be treated as patchy. However, there was little evidence that different species performed better on different types of patches. Instead, areas with deep or fine soils seemed to be favorable for a number of species. Results suggest that it is feasible to re-establish native trees and shrubs along large, regulated rivers, at least at certain sites for an initial period of several years with the aid of weed control and irrigation. Shallowly buried layers or lenses of gravel or sand are a hidden, fine-scale factor that can reduce plant growth on river terraces.