Vetiver buffer strips are widely employed to reduce fluxes of eroding soil and associated nutrients and pollutants from catchments into waterways. The physical processes involved in sediment reduction by buffer strips have been examined in a number of studies and various models have been developed to describe some of these processes. However, the sedimentremoval efficiency is complex as it is time-dependent and changes as deposition builds up and existing models do not allow for this, amongst other limitations. A new model, the Griffith University Soil Erosion & Deposition - Vegetative Buffer Strip Model (GUSED-VBS), has therefore been developed which couples the hydraulics, sediment deposition and subsequent adjustment to bed topography, to simulate the build-up of sediments in the backwater zone and its effect on flow conditions. The model can be used to predict sediment retention by buffers as well as sediment concentrations, size distributions and nutrient enrichment ratios of runoff. Experiments to characterize sediment and nutrient retention by a vetiver strip were carried out using surface flow in the Griffith University Tilting-Flume Simulated Rainfall facility and were later used to test the model. Replicate experiments were conducted at three slopes using a dense vetiver strip inserted into the flume. Water profiles were recorded, then sediment comprising either a sandy soil (Podzol), a red clay (Ferralsol) or a black clay (Vertisol) was introduced into flow upstream of the buffer and sediment deposition and outflow characteristics were measured for each soil. Total carbon, nitrogen and phosphorous levels were determined in different size fractions of each sediment, to characterize nutrient enrichment. Sediment loads in the outflow increased slightly with time for the Vertisol due to sediment movement into the buffer but were static for the other two soils. The buffering action of the vetiver was efficient, reducing sediment in the outflow to 3.2, 6.0 and 11.3 % of the inflow concentration for the Podzol, Ferrosol and Vertosol respectively, with a significant differences (P<0.01) between soils. Sediments in the runoff were primarily in the 0.002 - 0.2mm size range and the greatest enrichment of fines (silt size or smaller) occurred in the Ferralsol and Vertisol. Particulate-carbon, -nitrogen and -phosphorous levels in the outflow were reduced by more than 60% compared to the inflow. Measured data from the flume experiments were compared to simulated data from GUSED-VBS. Water profiles and total sediment deposition were simulated well with low root mean square errors and coefficients of efficiency approaching 1. Reductions of the inflow nutrient loads were fairly well simulated. Further work is underway to test the model using field data.