Phosphorus runoff research has been limited on soils that are calcareous to the surface. We followed the National P Runoff Protocol using rainfall simulation to develop a relationship between soil test P levels and runoff P losses on three Great Plains calcareous agricultural soils in Kansas, Nebraska, and Colorado. Each site had eight treatments with two replications. Manure was applied at 8 rates varying from 0 to 270 Mg ha-1. The rainfall simulation was applied at 70 mm hr-1 for one hour. Soil samples taken during characterization measured calcium carbonate concentrations varying from 1% to 9% by weight. The following soil sampling and analysis procedures were found to be optimum for estimation of the soil test P to runoff P relationship : soil sampling at 0-5 cm before rainfall simulation outside the sub-plot frame, using total dissolved phosphorus as the runoff analysis method and Mehlich-3 as the soil test phosphorus method. We also developed a multi-step regression analysis, which estimates total dissolved phosphorus concentration in runoff from the Mehlich-3 soil test P and the percent calcium carbonate by weight with an R-square value of 0.92. Phosphorus saturation was estimated for the three soils, and the relationship of the square root of the percent saturation as a function of the P added per treatment was determined. The Kim soil with the highest calcite concentration sorbed the most phosphorus; however, this resulted in the lowest percentage of P saturation in the soil due to the high Ca concentration. A soil with higher calcium carbonate percent by weight can sorb or be a sink for manure-P when applied to the soil. This ability to sorb P is only effective at reducing the total dissolved P in runoff, and the sediment P needs to be controlled with soil erosion practices.