Coastal stratigraphy from the Pacific Northwest of the United States contains evidence of sudden subsidence during ruptures of the Cascadia subduction zone. Transfer functions (empirical relationships between assemblages and elevation) can convert microfossil data into coastal subsidence estimates. Coseismic deformation models use the subsidence values to constrain earthquake magnitudes. To test the response of foraminifera, the accuracy of the transfer function method, and the presence of a pre-seismic signal, we simulated a great earthquake near Coos Bay, Oregon, by transplanting a bed of modern high salt-marsh sediment into the tidal flat, an elevation change that mimics a coseismic subsidence of 0.64 m. The transplanted bed was quickly buried by mud; after 12 mo and 5 yr, we sampled it for foraminifera. Reconstruction of the simulated coseismic subsidence using our transfer function was 0.61 m, nearly identical to the actual elevation change. Our transplant experiment, and additional analyses spanning the A.D. 1700 earthquake contact at the nearby Coquille River 15 km to the south, show that sediment mixing may explain assemblage changes previously interpreted as evidence of pre-seismic land-level change in Cascadia and elsewhere.