Phosphorus losses from arable soils contribute to eutrophication of freshwater systems. In addition to losses through surface runoff, leaching has lately gained increased attention as an important P transport pathway. Increased P levels in arable soils have highlighted the necessity of establishing a relationship between actual P leaching and soil P levels. In this study, we measured leaching of total phosphorus (TP) and dissolved reactive phosphorus (DRP) during three years in undisturbed soil columns of five soils. The soils were collected at sites, established between 1957 and 1966, included in a long-term Swedish fertility experiment with four P fertilization levels at each site. Total P losses varied between 0.03 and 1.09 kg ha(-1) yr(-1), but no general correlation could be found between P concentrations and soil test P (Olsen P and phosphorus content in ammonium lactate extract [P-AL]) or P sorption indices (single-point phosphorus sorption index [PSI] and P sorption saturation) of the topsoil. Instead, water transport mechanism through the soil and subsoil properties seemed to be more important for P leaching than soil test P value in the topsoil. In one soil, where preferential flow was the dominant water transport pathway, water and P bypassed the high sorption capacity of the subsoil, resulting in high losses. On the other hand, P leaching from some soils was low in spite of high P applications due to high P sorption capacity in the subsoil. Therefore, site-specific factors may serve as indicators for P leaching losses, but a single, general indicator for all soil types was not found in this study.