A study was conducted in Ihale Village, Magu District, Tanzania aimed to assess the effectiveness of selected vegetation cover types as Vegetative Buffer Strip Filters and their corresponding vegetation strip widths in filtering sediment delivery that are likely to enter the surface waters of the Lake Victoria from agricultural lands. Sediment laden runoff plots measuring 2 m by 20, 15, 12.5 and 10 m were established to determine the effectiveness of the selected filters and strip filter widths set at 10, 5 and 2.5 m against a standard width of 10 m which was planted with maize (Zea mays). The filters evaluated include elephant grass (Pennisetum sp) and Lemon grass (Cymbopogon citratus). Variables measured include: daily sediment for the rainy days, daily rainfall, infiltration rates, plant vigour of the filters (plant height) and percent vegetation cover. Data was analysed statistically using Excel, SAS and SPSS statistical packages. The results demonstrate that mass of sediment delivery through the tested filters decreases exponentially with increasing filter widths for all tested vegetation types with correlation coefficient (R 2 ) ranging from = 0.6 to 0.8. The trapping efficiency of the tested filters increases logarithmically with increasing filter widths. Sediment trapping efficiency increased from 54% at 2.5 m to 78% at 10 m vegetative filter widths for all tested vegetation types. A significant correlation exists between sediment delivery through elephant grass and lemon grass with filter strip width, percent vegetation cover and plant growth characteristics. About 30 to 40% of the observed variation in the prediction of sediment delivery through selected vegetation types could be explained by the percent vegetation cover alone. The study demonstrates that the tested Vegetative Buffer Strip Filters have the potential to purify water from sediment that has been transported into the riparian zone from the agricultural lands. Further research is recommended.