The cognitive radio technology is being developedin universities and industry as a solution to the radio spectrum scarcity. This technology willallow spectrum sharing between primary and secondary telecommunication users. The techniques employed must be robust enough to minimize spectrum occupancy collisions, when the dynamic spectrum access is applied. This study investigates the trends of spectrum usersoccupation, particularly in voice services in the 450 MHz frequency band.An users occupancy model was developed taking into accountmeasured data of real transmissions. It allowed the development of a robust algorithm that predicts spectral vacancy in channels allocated to primary users. The method selects, statistically, a group of future time intervals that can be used by secondary users, due to a higher probability of having a free spectral space. The use of this new technique minimizes possible collisions, increasing the flow of information in secure way and optimizing the radio spectrum use.