Soil phosphorus (P) fertility arising from historic P inputs is a major driver of P mobilisation in agricultural runoff and increases the risk of aquatic eutrophication. To determine the environmental benefit of lowering soil P fertility, a meta-analysis of the relationship between soil test P (measured as Olsen-P) and P concentrations in agricultural drainflow and surface runoff in mostly UK soils was undertaken in relation to current eutrophication control targets (30-35 mg P L À1 ). At agronomic-optimum Olsen P (16-25 mg kg À1 ), concentrations of soluble reactive P (SRP), total dissolved P (TDP), total P (TP) and sediment-P (SS-P) in runoff were predicted by linear regression analysis to vary between 24 and 183 mg L À1 , 38 and 315 mg L À1 , 0.2 and 9.6 mg L À1 , and 0.31 and 3.2 g kg À1 , respectively. Concentrations of SRP and TDP in runoff were much more sensitive to changes in Olsen-P than were TP and SS-P concentrations, which confirms that separate strategies are required for mitigating the mobilisation of dissolved and particulate P forms. As the main driver of eutrophication, SRP concentrations in runoff were reduced on average by 60 mg L À1 (71%) by lowering soil Olsen-P from optimum (25 mg kg À1 ) to 10 mg kg À1 . At Olsen-P concentrations below 12 mg kg À1 , dissolved hydrolysable P (largely organic) became the dominant form of soluble P transported. We concluded that maintaining agronomic-optimum Olsen-P could still pose a eutrophication risk, and that a greater research focus on reducing critical soil test P through innovative agro-engineering of soils, crops and fertilisers would give long-term benefits in reducing the endemic eutrophication risk arising from legacy soil P. Soil P testing should become compulsory in priority catchments suffering, or sensitive to, eutrophication to ensure soil P reserves are fully accounted for as part of good fertiliser and manure management.