Due to the implementation of an Adaptive Modulation-Coding (AMC) mechanism in the 802.16e physical layer, each connection's bandwidth requirements cannot be statically computed, but they derive as a function of the terminals' instantaneous Signal-to-Noise Ratio (SNR). Therefore, the Connection Admission Control (CAC) mechanism fails to establish an efficient policy that would optimally exploit the system's resources. In this respect, the present paper formulates initially an adequate method for statistically calculating the average capacity of a Mobile-WiMAX system, according to the distribution of the several attenuation factors that affect the signal along the propagation path. In parallel, based on the processing of the terminal's SNR samples, a sophisticated algorithm for filtering out any misleading measurements is developed. This second method aims at predicting the upcoming Modulation-Coding state of every connection, so as to acquire a short-term view of its spectrum demands. Finally, all the above info collected from both the estimation methods is utilized by the CAC procedure in order to perform an as accurate as possible computation of the resource availability. As a result, this hybrid approach succeeds in maximizing the network revenue through significantly increasing the number of concurrently serviced connections while guaranteeing their Quality of Service (QoS) standards.