Abstract Soils from a long-term slurry experiment established in 1970 at Hillsborough, Northern Ireland, were used in the experiment. The site has a clay loam soil overlying Silurian shale. Seven treatments were used with three replicate plots per treatment under the following manurial regimes: (1) mineral fertiliser supplying 200 kg N, 32 kg P and 160 kg K ha −1 yr −1; (2)–(4) pig slurry applied at 50, 100 or 200 m 3 ha −1 yr −1; (5)–(7) cow slurry applied at 50, 100 or 200 m 3 ha −1 yr −1. Agronomic measures of P determined on subsurface layers down to 90 cm were compared with sorption isotherm data and rates of desorption. Adsorption isotherms were fitted using a standard Langmuir model. Data were compared with soluble (molybdate-reactive) P levels in soil water collected at 35 and 90 cm using PTFE suction cup lysimeters. Agronomically available P was concentrated in the top 30 cm of soil in all treatments. The accumulation of P in surface layers of the plots was significantly greater in the pig slurry treatments compared to the cow slurry, reflecting the history of P amendments. Nevertheless, over a period of a year, molybdate-reactive phosphorus (MRP) concentrations in lysimeter collections was consistently higher at 35 cm depth in the highest cow slurry treatment (7) compared to the equivalent pig slurry treatment (4). Either the movement of soluble P down the profile is facilitated by the higher organic content of cow slurry or P movement is not directly related to P accumulation in the soils. In addition, it is hypothesised that P movement down the soil profile depends upon two separate mechanisms. First, a ‘break’ point above which the accumulated P in the surface horizons is less strongly held and therefore amenable to dissolution and movement down the profile. Second, a mechanism by which some solute P from the surface horizons can travel rapidly through horizons of low P status to greater depth in the soil, i.e., by preferential flow.