The long-term prediction of morphological bed evolution has been of interest to engineers and scientists for many decades. Usually, process-based models are employed to simulate bed-level changes in the scale of years to decades. To compensate for the major computational effort required by these models, various acceleration techniques have been developed, namely input-reduction, model-reduction and behaviour-oriented modelling. The present paper presents a new input-reduction method to obtain representative wave conditions based on the Shields criterion of incipient motion and subsequent calculation of the sediment pick-up rate. Elimination of waves unable to initiate sediment movement leads to additional reduction of model run-times. The proposed method was implemented in the sandy coastline adjusted to the port of Rethymno, Greece, and validated against two datasets consisting of 7 and 20 and 365 days, respectively, using the model MIKE21 Coupled Model FM. The method was compared with a well-established method of wave schematization and evaluation of the model’s skill deemed the simulations based on the pick-up rate schematization method as “excellent”. Additionally, a model run-time reduction of about 50% was observed, rendering this input-reduction method a valuable tool for the medium to long-term modelling of bed evolution.