Millions of dollars are spent annually on revitalizing salmon spawning in riverbeds where redd building by female salmon is inhibited by sediment that is too big for fish to move. Yet the conditions necessary for productive spawning remain unclear. There is no gauge for quantifying how grain size influences the reproductive potential of coarse-bedded rivers. Hence, managers lack a quantitative basis for optimizing spawning habitat restoration for reproductive value. To overcome this limitation, we studied spawning by Chinook, sockeye, and pink salmon (Oncorhynchus tshawytscha, O. nerka, and O. gorbuscha) in creeks and rivers of California and the Pacific Northwest. Our analysis shows that coarse substrates have been substantially undervalued as spawning habitat in previous work. We present a field-calibrated approach for estimating the number of redds and eggs a substrate can accommodate from measurements of grain size and fish length. Bigger fish can move larger sediment and thus use more riverbed area for spawning. They also tend to have higher fecundity, and so can deposit more eggs per redd. However, because redd area increases with fish length, the number of eggs a substrate can accommodate is maximized for moderate-sized fish. This previously unrecognized tradeoff raises the possibility that differences in grain size help regulate river-to-river differences in salmon size. Thus, population diversity and species resilience may be linked to lithologic, geomorphic, and climatic factors that determine grain size in rivers. Our approach provides a tool for managing grain-size distributions in support of optimal reproductive potential and species resilience.