In longitudinal studies it is often of interest to measure the association between a longitudinal marker repeatedly measured in time and the risk for an event. By postulating a model for the joint distribution, we acknowledge this link and we can assess the association between the longitudinal and the survival processes. In this work, we present the joint modeling approach and focus on the extension of the established methodology to include sequential survival times and more than one longitudinal variable. Thus, on the one hand, the purpose of the present work is to introduce the methodology for the joint models (comparing the frequentist and Bayesian approaches) and to develop a new theory to include more than one survival time and more than one longitudinal variable. On the other hand, we study in depth the structure of the R packages that fit joint models (JM and JMbayes) for one survival time and for one longitudinal variable in order to give some hints on how they should be modified to implement the problem we have theoretically formulated.