The resource-constrained nature of WSNs require efcient use of resources, especially energy, to prolong their lifetime. Clustering is one of the popular approaches to allocate the resources efciently among the WSN nodes. In this work, we analyze the problem of round length determination in cluster based WSN which has severe impact on the energy efciency. This problem is very important since round length determines how often the cluster head (CH) rotates or re-clustering process occurs. A longer round length will cause the CH nodes to operate for a long time and drain their energies faster than other nodes resulting in uneven energy consumption in the network, while a shorter round length results in considerable wastage of energy due to frequent running of the setup phase. Hence, we propose an adaptive and dynamic mechanism for round length determination in cluster based WSNs by adapting Behavior Curve Function modeled by quadratic Bezier curves, where we associate the remaining energy level of the cluster to its round operation length and to its assigned criticality which is defned based on network energy level. This helps to determine the number of frames in a round or how many times the data collection occurs in a cluster in a round and the criticality of the energy in the WSN. Simulation results reveal that the proposed mechanism has efectively reduced the energy consumption and improved the WSN lifetime in both homogeneous and heterogeneous network settings.