Considerable uncertainties are associated with the transfer of Tc from soil to plant. Mathematical models built to simulate that transfer generally use a single value derived from short-term experiments. Recent observations have, however, given a clear indication that the soil-to-plant transfer varies with time, depending on the biogeochemical cycling of this element. The long-term behavior of Tc in soils accidentally contaminated has been studied for 4 y in lysimeters exposed to natural climatic conditions. An important fraction of the deposit (70%) is rapidly removed from the 20-cm plow layer with a half-time of 54 +/- 2 d; the remaining fraction tends to persist in the soil for long periods of time (half-time = 30 +/- 16 y). Immediately after the deposit, the plant-to-soil concentration ratios are very high (about 400); they decrease to six 3 y after the contamination. The shape of the evolution of the contamination measured in grass after an accidental deposit is best-fitted to a sum of two exponential functions; the environmental half-times estimated for the two components are about 30 d and 2 to 3 y, respectively.