The aim of this thesis is to perform precise theoretical predictions of dark matter observables in the framework of different new physics models, and to compare these predictions with experimental data.

In a first step, radiative corrections applied to the direct detection of dark matter will be computed in the case of a fermionic, vector or scalar dark matter. In the last two cases, the results will be implemented in micrOMEGAs, a  software designed to predict, with high precision, consequences of new physics models for cosmology and direct or indirect detection of new particles at colliders  and other experiments. The implication of these precise computations for dark matter models with an extended Higgs sector and in extensions of minimal supersymmetric model will then be analysed.

In a second step, predictions for observables at colliders or in the flavor sector in the framework of these models will be improved in order to study the complementarity between different dark matter detection methods.
Forthcoming LHC data could influence the theoretical frame in which this second step will be performed.