Peatlands on which there is vacuum peat harvesting must be drained to allow peat extraction. Peat erosion due to mechanical tilling, vacuum harvest and rain can generate large suspended sediment concentrations that must be reduced before reaching the receiving waters. In this study, six sedimentation basins were monitored over one or two ice-free seasons, resulting in seven basin-years time series, to estimate their trapping efficiency. Using these data, two modelling approaches were first applied to predict the seasonal trapping efficiency of the basins, namely the Brown equation and a multilinear regression model (MLR). Secondly, an additional approach based on the hydrodynamic model MOHID was used to compute the trapping efficiency of one of the basins for a single rain event. The Brown equation, used worldwide in agricultural sedimentation ponds, proved to be less efficient in predicting the trapping efficiency of the basins (r = 0.67, p = 0.1, leave-one-out RMSE = 47.2%) than a MLR model using the volume of the basin, the surface of the harvested catchment and the Von Post decomposition degree of the peat as predictors (r = 0.87, p = 0.01, leave-one-out RMSE = 31.6%). The deterministic MOHID model was capable of reproducing high flow suspended sediment concentration at the basin outlet, based on measured settling velocity distribution of the peat through ViCAs protocol, but requires a good description of the settling velocity distribution across multiple hydrological conditions to simulate the suspended sediment concentrations over a full season in order to compute trapping efficiency. The results presented here provide new insights on eroded peat modelling and could lead to better design formulas for basin conception.