The potential of soil amendments in reducing phosphorus (P) loss has been reported under simulated snowmelt, but not evaluated in the field, under snowmelt conditions. We examined the effectiveness of alum (Al2(SO4)3·18H2O), gypsum (CaSO4·2H2O), and Epsom salt (MgSO4·7H2O) in reducing P loss from two flood-prone agricultural fields (non-manured field with initial soil test P concentrations, STP, of 76 mg kg−1 and manured field with initial STP of 202 mg kg−1), during snowmelt on Canadian Prairies. Amendments were applied (2.5 Mg ha−1) in 2020 fall to four replicated plots (3 m × 1 m) with pre-installed runoff boxes (1.2 m × 0.9 m). In the 2021 spring, snowmelt water in each runoff box was pumped out, volume recorded for each sampling day, and analysed for dissolved reactive P (DRP). Snowmelt DRP concentrations was greater in manured soil (0.73 ± 0.23 mg L−1) with high STP, than in non-manured soil (0.28 ± 0.10 mg L−1.) with low STP, implying a greater risk of P loss to snowmelt runoff when STP is high. Cumulative snowmelt DRP loads in non-manured soil ranged from 3.1 to 3.8 mg (equivalent to 0.03–0.04 kg ha−1) and were not significantly different among treatments. In the control treatment of the manured soil, the cumulative snowmelt DRP load from runoff boxes was 20.8 mg (equivalent to 0.19 kg ha−1), while amended treatments had 42 to 68 % lower cumulative DRP load than control, with the difference being significant only in Epsom salt treatment. Cumulative snowmelt DRP load was significantly and positively correlated to cumulative snowmelt volume (R2 = 0.84 for each soil). This study shows the potential of fall application of soil amendments, particularly Epsom salt, to reduce P loss to snowmelt from soils with high STP.