Two sequential extraction procedures were carried out on six soils with different chemical properties and contamination history to estimate the partitioning of U between different soil fractions. The first method (method of Schultz) was specifically developed for actinides, while the second one (method of Rauret) was initially created for heavy metals. A soil-to-plant transfer experiment was carried out with ryegrass to verify if one of the extracted fractions efficiently predicted plant uptake. In artificially contaminated soils, most of the U was retrieved from the exchangeable and the carbonates fractions. In soils with high natural levels of U or contaminated by industrial activity, most of the U was found in the less available fractions. The highest U transfer factors (TF) were observed for artificially contaminated soils and the lowest for soils with high natural concentrations or industrial contamination. The U concentration in the roots and shoots and the soil-to-root TF are well correlated to the U concentration determined in the first extracted fractions from the method of Schultz. Conclusion: the extraction method according to Schultz should be preferably used for U, and the exchangeable fraction can be proposed as a potential indicator to evaluate plant uptake in soils.