Flood-irrigation, to the extent that outwash (runoff from border-check bays) occurs, is a major cause of P loss from grazed pastures and has potential to harm surface water quality. We used a combination of rainfall simulation to produce runoff and field sampling of outwash to investigate processes of P loss from treatments receiving no irrigation and irrigation at 10%, 15%, and 20% soil moisture and every 21 days (3w). Intact soil blocks were removed from each treatment, dung removed, soils wetted to about 32% soil moisture, and runoff produced via rainfall simulation. This indicated that P losses were proportional to soil Olsen P concentrations (29.8–51.4 mg Olsen P/kg; 0.096–0.541 mg dissolved reactive P/L). Olsen P concentration was less in those treatments receiving a greater number of irrigations due to increased pasture production and, presumably, loss via outwash. When soil blocks were allowed to dry, concentrations in runoff more than doubled and were paralleled by a decrease in soil microbial biomass P. However, when outwash was sampled in the field, P loss was greater in more frequently irrigated treatments. This was attributed to increased stocking rates and P-release from dung masking any soil effect. However, differences in P loss in outwash from 2 treatments (without recent grazing) were attributed to different soil moisture before irrigation. Assuming 25% of irrigation is lost as outwash, annual loads were estimated to range from 0.7 kg P/ha in the 10% treatment, irrigated 2.6 times a year, to 12.6 kg P/ha in the 3w treatment, irrigated 6.5 times per year. This suggests that the frequency of irrigation and stocking rate dictate the majority of P lost in this system (not soil P concentration). Hence, mitigation practices should be promoted to minimise outwash in intensively sheep-grazed pastures and potential surface water quality impacts.