A large amount of information exists on the relations between Phosphorus (P) availability in soil and plant uptake whereas less information exists on the relations between P availability and the amount of P that can be lost to the environment. The potential of phosphorus release to subsurface drainage waters was evaluated with 3 unfertilized (0P) and 3 fertilized (P) agricultural soils using a flow-through reactor made of a column containing repacked soil leached with 2 mM CaCl2 for 28 days. Two sets of experiments were carried out. First, samples that had been incubated for 28 days in the presence of carrier-free 33P were introduced in the columns and the pH and the concentrations of 33P and P were measured in the eluates. In the second series, soils that had been incubated for 28 days without 33P were introduced in the columns and were removed after specific leaching times to assess P exchangeability with the isotopic exchange kinetics method. Between 1.3 and 32.6 mg P kg soil-1 and up to 39% of the added 33P were released from the columns after 28 days. 0P soils released smaller amounts of P and of 33P than the P soils. In two of the unfertilized soils, the release of 33P became non detectable after 48 hours of leaching. Two phases could be distinguished during the release of P from the columns. The first phase was marked by a significant pH decrease in the eluates and by high concentrations of P and 33P and lasted about 4 days. P release during this phase was probably resulting from a complex series of processes involving desorption/sorption and dissolution reactions. The second phase was characterized by stable pH values and by low rates of 33P and P release. P release during this phase resulted probably from desorption and diffusion processes. Results allowed measuring a D-value (mg P kg-1) which was the amount of isotopically exchangeable P having the same specific activity as the P released from the column. This amount was interpreted as the total amount of P that could be potentially released from the soil columns and varied between 172 and 284 mg P kg-1 in the soils where it could be calculated. Results obtained with the isotopic exchange kinetic experiment conducted on the soils that had been incubated without 33P and then leached with CaCl2 confirmed the complex processes that took place during the early days of the leaching experiment and showed that leaching strongly affected soil P exchangeability. Highly significant correlations were found between the amounts of CaCl2 extractable P measured in the soils before the leaching experiment and both the amounts of P released within 28 days from the columns and the amounts of P present in the rapidly leached pool. The interest and limits of this flow through reactor to assess P losses from soils were then discussed.