A method for computing phosphorus loadings from continuous stream discharge records and discrete water samples is presented. Because of the highly variable relationship between phosphorus concentrations and runoff discharges, loadings for runoff events with incomplete or no concentration data become difficult to compute. However, concentrations can be predicted from instantaneous discharge data using simple linear regression after separating the phosphorus data into the dissolved and particulate components. Dissolved phosphorus is stratified according to season when the ground is either frozen or not. Particulate phosphorus, because of its strong correlation to suspended sediment, is stratified according to the “event intensity” as determined by rainfall intensity and rate of rise of the stream hydrograph. The particulate data can be further stratified according to whether the sample was taken on the rise or fall (including peak) of the hydrograph. Field sampling recommendations are made based on the study findings.