Liquid swine manure was applied at 46,700 l ha^-1 during the winter to 4 small (approx. 1 ha) watersheds (WS) at the North Appalachian Experimental Watershed near Coshocton, OH.  In this 2-year study in east-central Ohio, the WS were in continuous no-till corn (Zea mays L.), and there were 30 m wide non-manured setbacks/filter strips downslope from each application area.  On two WS, the setbacks were corn stubble from the previous year while the other two WS had grass filter strips [orchardgrass, Dactylis glomerata L.; tall fescue, Schedonorus phoenix (Scop.) Holub, and oats, Avena sativa L.].  Three control WS, one with a corn setback and two with a grass filter, received mineral fertilizer during the growing season instead of manure during the winter.  Surface runoff was measured using H-flumes and sampled on an event basis using Coshocton wheels.  Within 60 days after manure application the first year, the grass filter allowed less transport of Total P (0.3 kg ha^-1) than the crop setback (2.4 kg ha^-1) (1.7 and 12.8 times the transport from the control WS, respectively).  Nitrate-N transport from the grass filter and crop setback was 0.7 and 1.7 kg ha^-1 (3.2 and 8.2 times greater than from the control WS), respectively.  When the time period was increased to include transport following application of mineral fertilizer (144 days following manure application), losses of P and N from the manured WS with grass filters were similar to or less than the transport from the control WS.  Differences between treatments were much smaller in the second year.  Even though winter application of manure carries a high risk of nutrient loss, grassed filter strips may reduce losses of N and P more than cropped setbacks.