Excessive or inappropriate application of surface applied P contributes to water quality impairment. There is insufficient information about short-term P loss with runoff for different P sources and timing of runoff events. Rainfall simulation and snowmelt collection techniques were used to study runoff P concentrations and loads when three P sources were applied after harvesting corn and soybean without tillage in 6 site-years. Diammonium phosphate fertilizer, liquid swine manure, and egg-layers manure were applied at 50 kg P ha^-1. Simulated rainfall was applied within 24 hr, after 10 days with or without light rainfall during the 10-day period, and in spring. Devices were installed to collect natural snowmelt. Runoff and snowmelt were analyzed for total P, bioavailable P, and dissolved reactive P. There were small or inconsistent differences between crop residues. Runoff P always was highest for fertilizer, and the ranking of swine and poultry manures changed across years and manure sources. A runoff delay significantly reduced runoff P from all sources, especially after light rainfall before the runoff event. Snowmelt P was very variable but was lowest for poultry manure. Soluble manure P did not explain differences between manure types. The physical form of the P source and how it reached soil under the residue often explained the differences. We conclude that P loss shortly after application to the soil surface is highest for fertilizer than for manure, but both P loss and differences among sources decreased significantly with a runoff delay. Therefore, the probability of an immediate runoff event plays a significant role in the risk of P loss from surface P application.