In this article, it is proposed the generalized beta model as a survival model to analyze the university dropout.This model captures some real characteristics of the data that not always are captured using the traditional models.The results are compared between two models; from the proposed model and from the traditionally used, lognormal model.It is possible to observe that there are situations in which the assumption of positive and unlimited distribution support is not adequate.This new approach contributes to the discussion about the university dropout.Furthermore this model can be converted into a new tool for the analysis about a current theme which is the dropout in the university population.