In this paper, we propose a random network to model the evolution of the academic performance focused on the educational level of Bachillerato in Spain. For that, we will build a random network. Once the random network is stated, it will evolve according to a set of evolution rules and this allows us to simulate the academic behavior of Bachillerato students in order to determine whether they promote or not to the next academic level. Finally, this simulation process will allow us to provide predictions given by 95% of confidence intervals of the Bachillerato academic results over the next few years. Our predictions provide good results since, for each academic subpopulation, the new real data appeared during the development of this study corresponding to the academic year 2009 − 2010 lie inside the confidence intervals.