Phosphorus (P) often limits the eutrophication of streams, rivers, and lakes receiving surface runoff. We evaluated the relationships among selected soil P availability indices and runoff P fractions where manure, whey, or commercial fertilizer applications had previously established a range of soil P availabilities on a Portneuf silt loam (coarse-silty, mixed, superactive, mesic Durinodic Xeric Haplocalcid) surface-irrigated with Snake River water. Water-soluble P, Olsen P (inorganic and organic P), and iron-oxide impregnated paper-extractable P (FeO-Ps) were determined on a 0.03-m soil sample taken from the bottom of each furrow before each irrigation in fall 1998 and spring 1999. Dissolved reactive phosphorus (DRP) in a 0.45-microm filtered runoff sample, and iron-oxide impregnated paper-extractable P (FeO-Pw), total P, and sediment in an unfiltered runoff sample were determined at selected intervals during a 4-h irrigation on 18.3-m field plots. The 1998 and 1999 data sets were combined because there were no significant differences. Flow-weighted average runoff DRP and FeO-Pw concentrations increased linearly as all three soil P test concentrations increased. The average runoff total P concentration was not related to any soil P test but was linearly related to sediment concentration. Stepwise regression selected the independent variables of sediment, soil lime concentration, and soil organic P extracted by the Olsen method as related to average runoff total P concentration. The average runoff total P concentration was 1.08 mg L(-1) at a soil Olsen P concentration of 10 mg kg(-1). Soil erosion control will be necessary to reduce P losses in surface irrigation runoff.