Phosphorus (P) loss from agricultural watersheds is generally greater in storm rather than base flow. Although fundamental to P-based risk assessment tools, few studies have quantified the effect of storm size on P loss. Thus, the loss of P as a function of flow type (base and storm flow) and size was quantified for a mixed-land use watershed (FD-36; 39.5 ha) from 1997 to 2006. Storm size was ranked by return period (<1, 1-3, 3-5, 5-10, and >10 yr), where increasing return period represents storms with greater peak and total flow. From 1997 to 2006, storm flow accounted for 32% of watershed discharge yet contributed 65% of dissolved reactive P (DP) (107 g ha(-1) yr(-1)) and 80% of total P (TP) exported (515 g ha(-1) yr(-1)). Of 248 storm flows during this period, 93% had a return period of <1 yr, contributing most of the 10-yr flow (6507 m(3) ha(-1); 63%) and export of DP (574 g ha(-1); 54%) and TP (2423 g ha(-1); 47%). Two 10-yr storms contributed 23% of P exported between 1997 and 2006. A significant increase in storm flow DP concentration with storm size (0.09-0.16 mg L(-1)) suggests that P release from soil and/or area of the watershed producing runoff increase with storm size. Thus, implementation of P-based Best Management Practice needs to consider what level of risk management is acceptable.