Dissolved inorganic P transport in runoff from agricultural soils is an environmental concern. Models are used to predict P transport but rarely simulate P in runoff from surface-applied manures. Using field-plot data, we tested a previously proposed model to predict manure P in runoff. We updated the model to include more data relating water to manure ratio to manure P released during water extractions. We verified that this update can predict P release from manure to rain using published data. We tested the updated model using field-plot and soil-box data from three manure runoff studies. The model accurately predicted runoff P for boxes, but underpredicted runoff P for plots. Underpredictions were caused by runoff to rain ratios used to distribute P into runoff or infiltration. We developed P distribution fractions from manure water extraction data to replace runoff to rain ratios. Calculating P distribution fractions requires knowing rainfall rate and times that runoff begins and rain stops. Using P distribution fractions gave accurate predictions of runoff P for soil boxes and field plots. We observed relationships between measured runoff to rain ratios and both P distribution fractions and a degree of error in original predictions, calculated as (measured runoff P/predicted runoff P). Using independent field-plot data, we verified that original underpredictions of manure runoff P can be improved by calculating P distribution fractions from measured runoff to rain ratios or adjusting runoff to rain ratios based on their degree of error. Future work should test the model at field or watershed scales and at longer time scales.