Maintenance of navigation channels in shelf seas with tidal sand waves usually requires repeated dredging operations. Optimizing these interventions is a difficult task, particularly complicated by the nonlinear morphodynamics of sand wave recovery after dredging. Here we present a process-based model study, incorporating different strategies of dredging in an existing nonlinear sand wave model. We consider ‘topping’ (removing sand from crest) and ‘swiping’ (same, but now placing it in the troughs), for a range of dredging depths. Starting point is a fully developed sand wave or sand wave field, as simulated in the model. Results indicate that sand wave recovery is slowest after swiping. Also, larger dredging volumes imply longer recovery times. Next, we study the maintenance intervals and (cumulative) dredging volumes resulting from adopting a typical temporal strategy: swipe to a prescribed dredging depth, as soon as the recovering sand wave crests have reached a critical depth. Maintenance intervals are found to depend on the dredging depth and, importantly, to vary over time, as well. This last result shows that sand wave recovery depends not only on height, but also on its shape, emphasizing the limitations of existing, empirical (Landau-type of) models based on amplitude (or height) only.