Historically in Puget Lowland rivers, wood jams were integral to maintaining an anastomosing channel pattern and a dynamic channel–floodplain connection; they also created deep pools. In the late 1800s, wood was removed from most rivers, rivers were isolated from their floodplains, and riparian forests were cut down, limiting wood recruitment. An exception to this history is an 11-km-long reach of the Nisqually River, which has natural banks and channel pattern and a mature floodplain forest. We use field and archival data from the Nisqually River to explore questions relevant to restoring large rivers in the Pacific Northwest and other forested temperate regions. In particular, we focus on the relation between recovery of in-channel wood accumulations and valley bottom forest conditions and explore implications for river restoration strategies. We find that restoring large rivers depends on establishing riparian forests that can provide wood large enough to function as key pieces in jams. Although the frequency of large trees in the Nisqually valley bottom in 2000 is comparable with that of 1873 land surveys, many formerly more abundant Thuja plicata (western red cedar) were cut down in the late 1800s, and now hardwoods, including Populus trichocarpa (black cottonwood) and Acer macrophyllum (bigleaf maple), are also abundant. Pseudotsuga menziesii (Douglas fir) and fast-growing P. trichocarpa commonly form key pieces that stabilize jams, suggesting that reforested floodplains can develop naturally recruited wood jams within 50 to 100 years, faster than generally assumed. Based on the dynamic between riparian forests, wood recruitment, and wood jams in the Nisqually River, we propose a planning framework for restoring self-sustaining dynamic river morphology and habitat to forested floodplain rivers.