Concern over nonpoint-source phosphorus (P) losses from agricultural lands to surface waters has resulted in scrutiny of factors affecting P loss potential. A rainfall simulation study was conducted to quantify the effects of alternative P sources (dairy manure, poultry manure, swine slurry, and diammonium phosphate), application methods, and initial soil P concentrations on runoff P losses from three acidic soils (Buchanan-Hartleton, Hagerstown, and Lewbeach). Low P (12 to 26 mg kg(-1) Mehlich-3 P) and high P (396 to 415 mg kg(-1) Mehlich-3 P) members of each soil were amended with 100 kg total P ha(-1) from each of the four P sources either by surface application or mixing, and subjected to simulated rainfall (70 mm h(-1) to produce 30 min runoff). Phosphorus losses from fertilizer and manure applied to the soil surface differed significantly by source, with dissolved reactive phosphorus (DRP) accounting for 64% of total phosphorus (TP) (versus 9% for the unamended soils). For manure amended soils, these losses were linearly related to water-soluble P concentration of manure (r2 = 0.86 for DRP, r2 = 0.78 for TP). Mixing the P sources into the soil significantly decreased P losses relative to surface P application, such that DRP losses from amended, mixed soils were not significantly different from the unamended soil. Results of this study can be applied to site assessment indices to quantify the potential for P loss from recently manured soils.