Abstract-In this paper, we address the problem of multiantenna spectrum sensing in Cognitive Radios (CRs) by considering the correlation between the received channels at different antennas. First, we derive the optimum genie-aided detector which assumes perfect knowledge of the antenna correlation coefficients, Primary User (PU) signal power and noise variance. This is used as a benchmark for comparing with more practical detectors when some or all of these parameters are unknown to the Secondary User (SU). Two scenarios are considered: 1) the antenna correlation coefficients and PU signal power are unknown to the SU; 2) the antenna correlation coefficients, PU signal power and noise variance are unknown to the SU. To derive sub-optimum detectors for these two scenarios, we apply the Rao test, an asymptotically equivalent test to the Generalized Likelihood Ratio Test (GLRT) that does not require the Maximum Likelihood (ML) estimates of unknown parameters. Additionally, we calculate analytical approximations to the detection and false-alarm probabilities of the proposed detectors and verify them with Monte-Carlo simulations. The simulation results show that these new detectors outperform several recently proposed detectors for CR using multiple antennas.Index Terms-Cognitive radio, spectrum sensing, multiple antennas, Rao test, antenna correlations, Fisher information matrix, noise variance mismatch, antenna array.