Plants release carbon‐based exudates from their roots into the rhizosphere to increase phosphorus (P) supply to the soil solution. However, if more P than required is brought into solution, additional P could be available for leaching from riparian soils. To investigate this further, soil columns containing a riparian arable and buffer strip soil, which differed in organic matter contents, were sown with three common agricultural and riparian grass species. The P loads in leachate were measured and compared with those from unplanted columns, which were 0.17 ± 0.01 and 0.89 ± 0.04 mg kg−1 for the arable and buffer strip soil, respectively. A mixture of ryegrass and red fescue significantly (p ≤ .05) increased dissolved inorganic P loads in leachate from the arable (0.23 ± 0.01 mg kg−1) and buffer strip soil (1.06 ± 0.05 mg kg−1), whereas barley significantly reduced P leaching from the buffer strip soil (0.53 ± 0.08 mg kg−1). This was dependent on the dissolved organic C released under different plant species and on interactions with soil management history and biogeochemical conditions, rather than on plant uptake of P and accumulation into biomass. This suggested that the amount and forms of P present in the soil and the ability of the plants to mobilize them could be key factors in determining how plants affect leaching of soil P. Selecting grass species for different stages of buffer strip development, basing species selection on root physiological traits, and correcting soil nutrient stoichiometry in riparian soils through vegetative mining could help to lower this contribution.