This thesis investigates one of the most important observables in cosmology: the galaxy two- point function. The proper characterisation of this quantity is of the utmost importance if we want to compare theoretical predictions with the experimental results coming from planned redshift surveys. In particular we focus here on the modification that light-cone effects in- troduce – at large scales – in the correlation function. We start by computing, through first-order cosmological perturbation theory, expressions describing the galaxy distribution in the universe, i.e. galaxy number counts. These results are then used in order to construct the fully relativistic correlation function, which is valid in full-sky, i.e. beyond the so-called flat-sky approximation. A publicly available code is provided for fast and accurate numerical evaluation of the two-point function which allows us to forecast the possibility, with future or planned experiments, of constraining cosmological parameters through the multipoles decom- position of the correlation function. Other relevant topics, related to the two-point function, that we discuss in this work are: the regularisation of an apparent infra-red divergence in the potentials terms such as the Sachs-Wolfe effect, the existence of a second feature in the matter correlation function due to the baryon-baryon correlation in the early universe and, finally, a framework to analyse a possible anisotropic component in the two-point function.