To prevent scalability problem caused by increasing traffic of real-time applications, cluster architecture with Time Division Multiple Access (TDMA) scheduling scheme is mostly deployed in wireless sensor networks. However, even though it have proven good scalability and suitability for real-time communications, but static scheduling lacks of adaptability in several situations by not admitting some real-time flows where slots remain available. To solve mentioned problem, in this paper, we propose an adaptive cluster-based scheduling scheme for real-time flows by utilizing the time slots for flows according to type of flows such as inside or outside cluster. So, the proposed scheme can achieve better utilization of channel by minimizing the number of unused channels for real-time flows than static scheme through new adaptive TDMA scheduling scheme. Simulation results show that more flows are admitted and delivered within the deadline in the proposed scheme by utilizing unused time slots accordingly.