An explosive growth in the number of mobile users necessitates higher capacities in the cellular network infrastructure. Emerging real-time services such as VoIP and Video on Demand have stringent delay requirements. Long Term Evolution (LTE) has been developed to provide higher data rates, lower latency for real-time services through Differentiated Services (DiffServ) support, and lower cost of operation with its all-IP based architecture. The LTE MAC scheduler plays a crucial role to achieve these objectives by distributing available radio resources among User Equipments (UEs). In this paper, a novel LTE downlink scheduling algorithm taking both channel qualities and traffic types of UEs into account is proposed. First, incoming traffic is classified into real-time and nonreal-time flows which are inserted into a priority list according to traffic types and channel qualities. The scheduler assigns available resource blocks to UEs using the order in the list. A reconflgurable parameter in the proposed scheduler determines the resource prioritization level for real-time services, and hence allows network administrators to more effectively control the resource allocation according to the QoS requirements of UEs and changing network dynamics. OMNeT++ based simulations demonstrate that the tradeoff between the performance improvement of UDP flows and the performance degradation of TCP flows can be controlled by tuning the reconflgurable traffic prioritization parameter of the proposed scheduling algorithm.