Our paper is focused on a subject having a great importance when Element Free Galerkin (EFG) method is used. Namely, it is about the choosing of some charateristic parameters of the kernel function. Practically, the EFG method user has to decide upon some aspects like which kernel function is going to be used, then which is the most fitted dimension of the domain support, which is the influence of the internodal distance for a better accuracy and others. This paper is based on the theoretical fundamentals of the EFG method and on our practical experience in using of the method. Our research presents some comments and recommendations accompanied by examples and processing of some results coming from practice. The author considers that despite all the progress made in the development of the EFG method, its use in the analysis of mechanical structures needs a good experience regarding some aspects on which the indications are few, or vague, or missing. The numerical examples in this work, are performing by using original programs in Matlab for solving simple problems. We consider that our conclusions are very useful not only for the beginners in using of EFGM but for those who want to know better and especially why it is better to use an one or other option.