This paper highlights the results obtained by designing a prediction model aimed at estimating the enrolment of students per subject in an institution of higher education from Mexico. To perform the estimation, the system dynamics technique was applied using software VenSim, version 7.2a. The model evaluated the period from August to December 2018 to forecast three subjects: Linear Algebra (AL), Applied Linear Programming (PLA) and Networks and Simulation (RYS). The model included the following variables: number of classrooms and teachers available; number of students who failed and passed in ordinary and extraordinary term; rate of student desertion. The main results showed a prediction accuracy percentage of 97% for AL, 89% for PLA and 97% for RYS. The prognostic error represented 3.4 groups in contrast to the prediction certainty of 72.3 groups. This research provides a model to forecast reliable data for decision making in order to optimize resources.