Most licensed spectrums in wireless communication systems are underutilized due to the fixed channel allocation policy. The limited availability of spectrum resource and great demand for spectrum access requires the deployment of techniques that enable efficient and timely utilization of available channels. With cognitive radio technology, secondary users (SUs) can use available spectrum holes of the primary users (PUs) without deteriorating the quality of service (QoS) of the PUs. Conventional spectrum holes detection and assignment methods consume a lot of time and energy power, which transpires to higher sensing error that increases spectrum handoff and negatively impacts data transmission. This paper develops a computational intelligence (CI)-based system that adaptively and dynamically manages spectrum resource in order to minimize spectrum handoff and sustain QoS. The developed system adopts Support Vector Machine (SVM) and Adaptive Neuro-Fuzzy Inference System (ANFIS) to predict spectrum holes availability and allocate best available channels to SUs. Then, a Java-based simulated network uses the output from the CI-based system and the interference temperature of selected channels in 2G, 3G and 4G spectrums to admit SUs’ requests for diverse service demands. Results indicate that SVM achieves 98.8% accuracy in spectrum holes detection while ANFIS achieves 90.4% accuracy in channel allocation. The prediction accuracy and generalization capability of these models demonstrate ability to achieve significant improvements in spectrum utilization. Network operators can leverage on this result to maximize throughput and maintain QoS and fairness among users.