Vegetated filter strips (VFS) have become important for water quality improvement. Locations to place VFS can be selected with Soil & Water Assessment Tool (SWAT), a comprehensive watershed-scale water quality model. However, the VFS efficiency cannot be estimated with SWAT because the VFS factors, e.g. vegetation status, soil infiltration capacity, and rainfall intensity and duration operate at much finer time scales than the SWAT simulates. Our objective was to develop a model for transport of manure-borne E. coli in VFS, to allow the estimation of the uncertainty in its parameters, and to evaluate the effect of the parameters uncertainty on the uncertainty in simulated VFS efficiency. We developed a physically-based model (STIR, Solute Transport with Infiltration and Runoff). The model can simulate the overland transport and loss to infiltration of manure-borne E. coli. and has been parameterized and tested using experimental data on release and transport of manure borne E. coli under simulated rainfall at the ARS experimental site in Beltsville, MD. The uncertainty in transport parameters was estimated from this dataset, the uncertainty in soil hydraulic properties was estimated from the UNSODA database, the climatic uncertainty was assessed from historic records, and the uncertainty in surface runoff parameters was evaluated from literature. Then Monte Carlo simulations were applied to evaluate the variability in E. coli breakthrough in VFS. Overall, for the example of our experimental 6-m long VFS at the 20% slope, the VFS efficiency was less than 100% and less than 75% in 5% and in 2.5% of cases, respectively. Specific factors caused the VFS partial failure will be discussed. The simulations showed that probabilistic characterization of the VFS efficiency with site-specific soil, vegetation, and weather properties is possible and may be useful in making decisions on VFS placement with respect to manure-borne pathogen microorganisms.