Understanding phosphorus (P) accumulation and stratification in surface soil following manure application is critical for runoff P loss abatement. The objectives of this study were to predict soil P stratification and assess its effects on runoff P losses following long-term manure application. A two-step approach was developed to use field data and predict P stratification and runoff over time. First, extractable soil P in 0 to 5, 5 to 10, and 10 to 15-cm (0 to 2, 2 to 4, and 4 to 6-in) depths were measured in a 2.5 ha (6.2 ac) watershed, where cattle and swine manure were applied for four consecutive years. Second, the measured P and management practices were used to calibrate input parameters of the Groundwater Loading Effects of Agricultural Management System (GLEAMS) simulation. The calibrated parameters were used to estimate soil P stratification and runoff P losses following 25 years of simulated annual manure application. The Iowa P index rating was calculated for the same scenarios. Correlation among extractable P fractions were high (r(2) > 0.88). The GLEAMS simulation predicted reasonably well the exacerbation of soil P stratification following manure application. The simulation estimated a 5.3 mg L-1 and 7.4 mg L-1 runoff P increase for each 1 mg kg(-1) soil labile-P increase in the 0 to 1-cm and 1 to 5-cm depths (0.4 and 2 in), respectively. Soil Bray-P in 0 to 5-cm vs. 0 to 15-cm depth increases P-index risk towards manure application rates. Prediction of P stratification with models is useful for improving P loss risk assessment when combined with the P-index and thus provisions should be made for P stratification in comprehensive nutrient management programs.