Pasture-based grazing systems contribute to the excessive nutrients found in some streams in southern Australia. This study investigated P exported from a 3.6 ha dairy pasture receiving high levels of P fertilizer with a view to identifying factors affecting P export. The runoff from 34 storms was monitored to examine factors thought to be associated with the variation in total phosphorus (TP). The relationships between TP and days since grazing (DG), days since fertilizing (DF), total storm flow (TF), and season (year and time of year) were examined. The variable DF was inversely related to the flow-weighted mean TP concentration (P < 0.001) and accounted for 50.6% (adjusted R 2 ) of the variance in ln (TP). However, DG was only weakly correlated to TP, suggesting that cattle did not mobilize large stores of available P relative to the fertilization at the study site. The initial half-life of fertilizer effect (i.e. the period of time since fertilizing in which the TP concentration decreased by half) was estimated to be 4.3 d, with a 95% confidence interval of 3.1 to 8.0 d. However, unlike exponential decay where the half-life is constant, the decay rate for these data declined over time. The data adjusted for the effects of DG and TF show little evidence of decline in TP after 20 d since fertilizing. Management of fertilizer application in relation to the timing of runoff events appears the main method by which P export can be decreased in the type of pasture-based grazing system studied here.