The risk of P loss from manured soils is more related to P fractions than total P concentration in manure. This study examined the impact of manure P fractions on P losses from liquid swine manure- (LSM), solid cattle manure- (SCM), and monoammonium phosphate- (MAP) treated soils. Manure or fertilizer was applied at 50 mg P kg(-1) soil, mixed, and incubated at 20 degrees C for 6 wk to simulate the interaction between applied P and soil when P is applied well in advance of a high risk period for runoff: Phosphorus fractions in manure were determined using the modified Hedley fractionation scheme. We used simulated rainfall (75 mm h(-1) for 1 h) to quantify P losses in runoff from two soils (sand and clay loans). The proportion of total labile P (total P in water+NaHCO3, fractions) in manure was significantly greater in LSM (70%) than SCM (44%). Mean dissolved reactive P (DRP) load in runoff over 60 min was greatest from MAP-treated soil (18.1 mg tray(-1)), followed by LSM- (14.0 mg tray(-1)) and SCM- (11.0 mg tray(-1)) treated soils, all of which were greater than mean DRP load from the check (5.2 mg tray(-1)). Total labile P (water+NaHCO3) in manure was a more accurate predictor of runoff DRP loads than water extractable P, alone, for these two soils. Therefore, NaHCO3, extraction of manure P may be a useful tool for managing the risk of manure P runoff losses when manure is applied outside a high risk period for runoff loss.