The response of an ocean bottom seismometer (OBS) to a transient pull that excites the natural OBS-sediment coupling resonance can be modeled as a mass-spring-dashpot system in which the resonant frequency and damping are functions of instrument mass and bearing radius and of the physical properties of the sediment (primarily the shear modulus). For the very soft sediments sometimes found on the sea floor, this resonance may be within the main frequency band of interest (2 to 15 Hz) for many common instrument configurations. To test the model and to find an anchor that would shift the coupling resonance to a higher frequency and decrease its amplitude, we conducted a series of tests which measured the response of the vertical and horizontal components of the U.S. Geological Survey OBS to transient pulls as a function of anchor configuration and sediment properties. The tested anchors included a concrete “flowerpot,” a tripod, a plate, and a perforated plate. Sites were on soft, organic-rich ooze and on firm sand. Several small shots were also fired at the ooze site in order to compare the response of the plate and “flowerpot” anchors to seismic signals. For a given anchor at a given site, the observed response was very repeatable. We found that the model predicts the vertical coupling response quite well and that good vertical coupling can be achieved with the plate or perforated-plate anchors. The response to the horizontal pulls, however, was similar and resonant for all anchors.