A key issue faced in dam removal is the rate and timing of remobilization and discharge of stored reservoir sediments following the removal. Different removal strategies can result in different trajectories of upstream sediment transport and knickpoint migration. We examine this issue of for the Marmot Dam removal in Sandy River, Oregon, USA using both physical experiments and field studies accompanying removal of the dam in October 2007. The physical experiment was designed to provide insights on how and if the position of a cofferdam notch will affect how reservoir sediments are remobilized, with the goal of minimizing the volume of sediment stranded in terraces. Data and observations indicate that at lower failure discharges, notch position impacts the location of cofferdam failure as well as the location of the first major knickpoint and its trajectory. In particular, notch positions that force the river to migrate laterally in order to adjust to natural valley orientation and morphology were most effective in removing larger volumes of sediment and reducing terrace heights. Actual cofferdam notching to maximize erosion produced extremely rapid and significant erosion of reservoir sediments. Comparison of model results with field observations suggests that the physical experiments provided solid predictions of rates of erosion and overall knickpoint trajectory.