With the technological development of the telecommunications industry, the radio spectrum is almost fully occupied with a large number of multiple assignments for wireless services for both commercial and non-commercial applications, such as defense, air traffic control and scientific exploration. The electromagnetic spectrum is a precious and scarce natural resource. Therefore, a considerable effort is being directed at the development of cognitive radios, capable of sensoring the spectrum and using momentarily available frequency bands in an opportunistic way. The tracking and using of these spectral intervals, or white spaces, using cognitive radio technology will enhance the efficiency of the spectrum use and allow the introduction of new telecommunications services to be exploited by secondary users, obliged not to interfere or produce very limited interference to primary users. The aim of this study is to evaluate the main algorithms for detection of spectral intervals (Energy Detector, Detection of Covariance Absolute Value, Spectral Covariance Sensing) through simulations with experimental data obtained in field measurements campaigns. The algorithms were tested to evaluate their performance in terms of detection probability given a required false alarm probability, computational complexity and robustness in low signal-to-noise conditions. The experimental data used comes from the measurements campaigns in urban environments at the 3.5 GHz band.