Physics of the structure of liquid metals boasts about a double diversity. Firstly, numerous potentials exist to describe the interactions between particles. Secondly, a large number of approaches have been proposed to deduce the structure from the effective potential. In this work, we study the structure of liquid alkali metals. It is developed around two ideas, attributing a central role to molecular dynamics results. In a first part, the quality of the potential implemented is discussed through a comparison between experimental and simulation results. We used Shaw's model potential and both the Vashishta-Singwi and the Ichimaru-Utsumi local field corrections. It appears that this ab initio potential describes correctly the structure of each alkali metal, including lithium. Molecular dynamic's results seem quite insensitive to the choice of the dielectric function. This confirms the predominant role played by short range forces in determining the structure. The second idea, this study is built on, is an evaluation of different methods available for the description of the structure. By comparison with molecular dynamics, qualities and defaults of both perturbation schemes (ORPA-WCA, ORPA-JA) and integral equations (SMSA) are discussed. In the cases of Na, K, Rb and Cs, these methods produce results near simulation ones. However, the SMSA equation does not show the characteristic drawbacks of perturbation methods. Lithium is particular since any of these analytical methods achieves in matching, even approximately, simulation results. The reasons are not clearly understood. Screening influences S(o) and we underline that its value depends on the way long range interactions are taken into account