This paper investigates the performance of iterative interference alignment (IA) with spatial hole sensing in K-user multi-input multi-output (MIMO) cognitive radio (CR) networks. In the considered network, there are some unused degrees of freedom (DoF) or equivalently spatial holes in the primary network (PN) where the secondary network (SN) users communicate without causing harmful interference to the PN receivers. First, the generalized likelihood ratio test method is utilized to determine the availability of the unused DoFs; then, it is decided whether individual primary streams are present in the PN. With the aid of precoding and suppression matrices generated by an iterative IA approach, the interferences in the PN that are caused by the SN are aligned, and due to the secondary transmission, interference leakage on the kth primary receiver decreases below 10−6. The effects of the detection threshold values and the number of transmitter and receiver antennas are investigated in terms of detection and false alarm probability. Finally, the amplify-and-forward (AF) relaying scheme in the SN is evaluated and the impact of the relaying architecture on the system performance is analyzed.