This study investigated the effectiveness of vegetated buffer strips for removing contaminants in runoff from grassed plots (slope 15%) after application of cattle slurry. Plots (8 x 8 in 2 or 8 x 3 in 2) received slurry or inorganic fertilizer, and then simulated rainfall (1, 7 and 21 days after slurry/fertilizer application), after each event, runoff and percolates were sampled at various distances downslope (2,4,6, and 8 in), and analysed for Cl-, Na+, K+ and faecal bacteria contents. Contaminant concentrations were markedly higher in runoff from the slurry-amended plots than in runoff from the fertiliser-amended plots. After the first rainfall event, some contaminant concentrations in runoff from the slurry-amended plots declined with distance downslope (i.e. with buffer strip width), supporting the relative efficacy of the strip for retaining pollutants. After the second and third rainfall events, by contrast, our results suggest remobilisation of contaminants retained during the first event. Faecal bacteria levels (especially streptococcus levels) remained high throughout the study, even in percolates and runoff collected 8 m downslope after the third rainfall event, and indeed even downslope of the adjacent fertilizer-amended plots (indicating lateral movement): this suggests that bacterial contamination may be the most significant risk arising from slurry application.