This paper is focused on increasing the knowledge on methods for calibrating BES models and to get more insights of different approaches for the optimization of the calibration process. The paper will be centred in the evaluation of a multistage guided search approach. It defines an iterative optimization procedure which starts with the assignment of probabilistic density functions to the unknown parameters, followed by a random sampling and running batch of simulations. It then finishes with an iterative uncertainty and sensitivity analysis combined with a re-assignment of the ranges of variation of the strong parameters. The procedure converges when no new influencing parameters are found. This method is applied to a real case study consisting of an unoccupied office building located in Lleida (Spain). The measured indoor temperature has been used to determine the uncertainty and precision of the method. The effect of the size of the sampling, the number of iterations and the parameters of the global sensitivity method are analyzed in detail. The results of this paper exemplify the degree of accuracy of multistage guided search approaches, and illustrate the reasons how these analyses can contribute to the improvement of more refined calibration methods.