Phosphorus (P) released from agricultural soils to surface flood water and soil pore water during flooding contributes to P loading in watersheds and eutrophication in lakes. Identifying soils with a high risk of P losses to flood water is important in selecting proper management practices to minimize P losses. This study examined the effectiveness of various soil P extraction methods in predicting the potential P release to surface flood and soil pore water. Twelve alkaline to calcareous soils from across Manitoba were subjected to two treatments: an unamended control and solid dairy cattle manure amended at the rate of 100 kg P ha^-1, and were incubated aerobically at room temperature for four weeks prior to flooding. After incubation and prior to flooding, soil subsamples were analyzed for extractable P using the Olsen (Ols-P), Mehlich-3 (extractable P measured as molybdate reactive P (M3P_MRP) and by ICP-AES (M3P_ICP)), water (WEP) and modified Hedley fractionation methods. Soils were flooded for 56 days at room temperature.  During the flooding period, dissolved reactive P (DRP) was measured in surface flood water and soil pore water on a weekly basis. Release of P into surface and pore water was highly variable among soils depending on the soil properties. According to regression analysis, M3P_ICP (r²=0.67 to 0.89) followed by WEP (r²=0.41 to 0.85) were the most reliable P extraction methods for predicting the DRP concentration in surface and pore water three weeks after flooding. The M3P_MRP and Ols-P extraction methods were also good predictors but did not predict DRP concentrations as reliably as the other two methods, especially for manured soils. For the fractionation analysis, Ca-bound P which was extracted by 0.1 M HCl was most consistently related with DRP concentration in surface and pore water (r²=0.26 to 0.67).