A potential climate change might also bring changes in the extreme characteristics of most climatic variables, particularly on rainfall. Modeling extreme rainfall behavior is important due to the impact of natural hazards on highly vulnerable zones. In this regard, and within the framework of the classical extreme theory, the Generalized Extreme Value (GEV) model is proposed for the study of the behavior of extreme rainfall events in Venezuela. A Bayesian approach was used to estimate model parameters and to make predictive inference of the GEV model. Markov Chain Monte Carlo (MCMC) methods were used to get samples from the posterior distributions of the GEV model parameters. Numerical results are presented for six locations in Venezuela representing different mesoclimate types: La Mariposa (Miranda State); San Francisco de Macanao (Nueva Esparta State); Villa El Rosario (Zulia State); Machiques (Zulia State); Carora (Lara State); and San Carlos de R�Negro (Amazonas State). Simulations from the predictive distribution suggest that the Fr 飨et and Gumbel models are more appropriate to represent the annual maxima in most of the study locations; however, in locations with extreme conditions within arid or highly humid mesoclimates, the Weibull model is more appropriate to represent annual rainfall maxima.