Vegetative filter strips (VFS) are areas of either planted or indigenous established vegetation designed to improve the quality of surface runoff. They can be incorporated into pastures, grassed waterways, terraces, or cropland to remove sediment, nitrogen, or phosphorus from runoff. However, their trapping efficiencies depend mainly on the sediment characteristics. The finer soil particles pass through the filter more easily and have more surface area, accounting for most of the transported sediment-bound pollutants. Therefore, the objective of this study was to use a modeling approach to develop a Sediment Composition Vegetative Filter Strip (SCVFS) model to predict the clay, silt and sand trapping efficiencies of VFS based on a wide range of soil loss data from Central Chile. The physically based Water Erosion Prediction Project (WEPP) model was implemented using actual soil and climate data from Central Chile to generate a soil loss database for many VFS designs. More than 22,000 erosion events were generated with data from 28 sites. Three nonlinear relationships were developed between clay, silt and sand delivery and VFS length. The clay and silt models provided accurate estimates compared to WEPP for the calibration (R2=0.80–0.87) and validation (R2=0.73–0.80) sites. However, the sand estimates were not correlated to WEPP (R2=0.00–0.18), as the model computes sand using the difference between the total predicted sediments (R2=0.77–0.89) and the clay and silt estimates. Nevertheless, because most of the sediment-bound pollutants move with the finer sediments, this is not a modeling limitation. Compared to other physically based VFS models, the SCVFS model is remarkably easy to use because it only requires the characteristics of the sediment delivery when there is no filter, the rainfall erosivity and the hillslope and filter lengths. Additionally, it can be easily combined with many existing erosion models that provide daily soil loss estimates, making it a flexible VFS design tool. This study provides the detailed methodology to construct this model and discusses its advantages and limitations.