Résumé : We improve standard big bang nucleosynthesis (SBBN) calculations by taking into account new nuclear physics analyses (the 2003 work of Descouvemont and coworkers). Using a Monte Carlo technique, we calculate the abundances of light nuclei (D, 3He, 4He, and 7Li) versus the baryon-to-photon ratio. The results concerning Ωbh 2are compared with relevant astrophysical and cosmological observations: the abundance determinations in primitive media and the results from cosmic microwave background (CMB) experiments, especially the Wilkinson Microwave Anisotropy Probe (WMAP) mission. Consistency between WMAP, SBBN results, and D/H data strengthens the deduced baryon density and has interesting consequences on cosmic chemical evolution. A significant discrepancy between the calculated 7Li abundance deduced from WMAP and the Spite plateau is clearly revealed. To explain this discrepancy, three possibilities are invoked: systematic uncertainties on the Li abundance, surface alteration of Li in the course of stellar evolution, or poor knowledge of the reaction rates related to 7Be destruction. In particular, the possible role of the up to now neglected 7Be(d, p)2 α and 7Be(d, α)5Li reactions is considered. Another way to reconcile these results coming from different horizons consists of invoking new, speculative primordial physics that could modify the nucleosynthesis emerging from the big bang and perhaps the CMB physics itself. The impressive advances in CMB observations provide a strong motivation for more efforts in experimental nuclear physics and high-quality spectroscopy to keep SBBN in pace.