Time-dependencies of suspended-sediment discharge from six coastal watersheds of northern California – Smith River, Klamath River, Trinity River, Redwood Creek, Mad River, and Eel River – were evaluated using monitoring data from 1955 to 2010. Suspended-sediment concentrations revealed time-dependent hysteresis and multi-year trends. The multi-year trends had two primary patterns relative to river discharge: (i) increases in concentration resulting from both land clearing from logging and the flood of record during December 1964 (water year 1965), and (ii) continual decreases in concentration during the decades following this flood. Data from the Eel River revealed that changes in suspended-sediment concentrations occurred for all grain-size fractions, but were most pronounced for the sand fraction. Because of these changes, the use of bulk discharge-concentration relationships (i.e., “sediment rating curves”) without time-dependencies in these relationships resulted in substantial errors in sediment load estimates, including 2.5-fold over-prediction of Eel River sediment loads since 1979. We conclude that sediment discharge and sediment discharge relationships (such as sediment rating curves) from these coastal rivers have varied substantially with time in response to land use and climate. Thus, the use of historical river sediment data and sediment rating curves without considerations for time-dependent trends may result in significant errors in sediment yield estimates from the globally-important steep, small watersheds.