Phosphorus (P) concentration and flow data gathered during the 1990s for a range of British rivers were used to determine the relative contributions of point and diffuse inputs to the total P load, using the Load Apportionment Model (LAM). Heavily urbanised catchments were dominated by sewage inputs, but the majority of the study catchments received most of their annual phosphorus load from diffuse sources. Despite this, almost 80% of the study sites were dominated by point source inputs for the majority of the year, particularly during summer periods when eutrophication risk is greatest. This highlights the need to reduce sewage P inputs to improve the ecological status of British rivers. These modelled source apportionment estimates were validated against land-use data and boron load (a chemical marker for sewage). The LAM was applied to river flow data in subsequent years, to give predicted P concentrations (assuming no change in P source inputs), and these estimates were compared with observed concentration data. This showed that there had been significant reductions in P concentration in the River Thames, Aire and Ouse in the period 1999 to 2002, which were attributable to the introduction of P stripping at sewage treatment works (STW). The model was then used to forecast P concentrations resulting from the introduction of P removal at STW to a 2 or 1 mg l− 1 consent limit. For the urbanised rivers in this study, the introduction of phosphorus stripping to a 1 mg l− 1 consent level at all STW in the catchment would not reduce P concentrations in the rivers to potentially limiting concentrations. Therefore, further sewage P stripping will be required to comply with the Water Framework Directive. Diffuse P inputs may also need to be reduced before some of the highly nutrient-enriched rivers achieve good ecological status.