The structure of natural uranium-bearing, recently formed products arising from oxidative weathering of a U deposit (Peny, Massif Central, France) was studied to determine mechanisms of U-trapping by natural gels. Sampled waters are close to saturation with respect to amorphous silica and crystalline-hydrated U-hydroxides and silicates. All products are U-bearing Si/Al- and Fe-rich gels with minor Ca, Mg and water. Fourier Transform InfraRed spectrometry (FTIR) and X-ray absorption spectroscopy at Al-K, Fe-K and U-L I I I edges show that these gels are composed of intimate mixtures of short-range ordered aluminosilicates and hydrous ferric oxides. In Si/Al-rich gels, Al is 6-fold coordinated to oxygens and Si tetrahedra are mostly isolated as in Al-rich allophane (Al:Si=2). Fe-rich gels exhibit a local structure dominated by edge-sharing octahedra, with d(Fe-Fe)=3.03 9, likely resulting from hydrolysis and oxidation from Fe 2 + solution, and poisoning by Si/Al during Fe 3 + precipitation. The local Al environment is close to that measured in Al-substituted goethite, and the Si tetrahedra are poorly polymerized as in natural Si-bearing ferrihydrite. The local structure around U was solved up to 3-4 9 by Extended X-ray Absorption Fine Structure (EXAFS). Uranium is present in the hexavalent oxidized state, as uranyl complexes (UO 2 2 + ) characterized by two axial oxygen atoms at d(U-O a x )=1.80 9 and four equatorial oxygen atoms, at two distinct distances (d(U-O e q 1 =2.33 9, d(U-O e q 2 )=2.48 9). The collinear O a x -U-O a x geometry explains the presence of multiple scattering (MS) events in the EXAFS spectra of the U-bearing gels. In Si/Al-rich gels, U-U pairings at 3.82 9 are consistent with polymers based on edge-sharing uranyls, in accordance to the speciation calculated in associated solutions. Alternatively to U-O a x MS, the contribution of Si neighbors at 3.3 and 3.7 9 accounts for structural data beyond uranyl polyhedra. It is consistent with the local structure of uranophane, suggesting a coprecipitation process of U and Si. Elements such as Al may have poisoned crystal growth. In contrast, no U-Fe, U-Si/Al nor U-U contributions are evidenced in the Fe-rich gels. Only a weak contribution due to U-O a x MS is present. The geometry of O e q coordination shell suggests that uranyl is complexed or sorbed onto mineral surfaces. It is proposed that U has been complexed by low Z elements like Si, Al in a first step, then trapped within hydrous ferric oxides during iron precipitation in a second step, i.e., during the final oxidation of the solutions inside the mine galleries.