The acceleration of surface water eutrophication attributed to agricultural runoff has focused attention on manure management in no-till. We evaluated losses of phosphorus (P) in sub-surface and surface flow as a function of dairy manure application to no-till soils in north-central Pennsylvania. Monitoring of a perennial spring over 36 months revealed that dissolved reactive P (DRP) concentrations increased 3- to 28-fold above background levels whenever manure was broadcast to nearby field soils. A study conducted with 30-cm deep intact soil cores indicated that incorporation of manure by tillage lowered P loss in leachate relative to broadcast application, presumably due to the destruction of preferential flow pathways. More P was leached from a sandy loam than a clay loam soil, although differences between soils were not as great as differences between application methods. In contrast, rainfall simulations on 2-m2 field runoff plots showed that total P (TP) losses in surface runoff differed significantly by soil but not by application method. Forms of P in surface runoff did change with application method, with DRP accounting for 87 and 24% of TP from broadcast and tilled treatments, respectively. Losses of TP in leachate from manured columns over 7 weeks (0.22–0.38 kg P ha−1) were considerably lower than losses in surface runoff from manured plots subjected to a single simulated rainfall event (0.31–2.07 kg TP ha−1). Results confirm the near-term benefits of incorporating manure by tillage to protect groundwater quality, but suggest that for surface water quality, avoiding soils prone to runoff is more important.