In this paper we discuss some aspects regarding the role of the guiding function and the optimization process in obtaining vibrational spectra of molecules using correlation function quantum Monte Carlo method. We start with the guiding function most used in these calculations and verify how its parameters affect the integration of the matrix elements involved in the problem. We then study how the optimization process influences the accuracy of the results. The results point out that there is an optimum parameterization of the guiding function, which in turn gives better accuracy and lower statistical errors. They also indicate that in order to improve the accuracy of the higher excited states, one should include them in the optimization process.