Abstract : Atmospheric neutrinos are produced by interactions of the cosmic rays with the atmosphere's nuclei. The detection of these neutrinos on Earth is currently achieved with large underground water Cerenkov detectors. These experimental observations are compatible with a deficit of detected muonic neutrinos in comparison with the theoretical predictions. Thus the observed ratio of muonic to electronic neutrinos is smaller than the theoretical one (up to a factor 2) and this constitutes the so-called atmospheric neutrino anomaly. This anomaly could be linked to that observed in the solar neutrinos experiments. If so, interpretations in terms of neutrino oscillations on their way from the place where they are produced and the place where they are detected could be attractive explanations. These oscillation phenomena assume the existence of massive neutrinos and therefore require the introduction of new theoretical mechanisms beyond the Minimal Standard Model. The aim of this work is to evaluate the effects of nuclear correlations on the interactions between the atmospheric neutrinos and the oxygen nuclei of the water Cerenkov detectors. The products of these interactions, if emitted above the Cerenkov threshold, are detected and identified thanks to the Cerenkov light ring they produce. The events are then classified according to the number of produced rings. It is therefore absolutely necessary to compute the neutrino-oxygen events rates in each exclusive reaction channel. The interpretation of the experimental results has been up to now limited to the quasi-elastic nucleon and Delta channels. Beyond this approximation, there exists other reaction channels which are able to lead to misidentification problems. Among these the non pionic disintegration channels of the Delta resonance play a special role in the sense that they induce single ring events that have not been considered up to now. To perform such calculations we adopt the nuclear response formalism. Our approach starts with a semi-classical approximation. This allows us to include in an economic way the effects of the nuclear correlations by solving exactlty the RPA equations in the ring approximation. The results we have obtained show that the nuclear correlations strongly modify the inclusive and exclusive neutrino-oxygen cross-sections and absolute interaction rates and that the ratio of the interaction rates mu/e is not very much affected by these nuclear effects. The analysis in the exclusive channels leads to the result that the number of pions predicted in the simulations is overestimated. In conclusion this work has shown the importance of the nuclear correlations in the neutrino-oxygen interaction and its impact on the atmospheric neutrino anomaly. It goes beyond the usual quasi-elastic approximations and can be moreover extended to others target-nuclei, such as iron, already used in past experiments. It is an interesting and predicitive tool for future neutrinos experiments based on differents kinds of nuclear targets.