Considering both channel quality and stability may enhance the likelihood of optimum channel decision for stream-based communication. Good quality stable channels may maintain a particular quality level during data stream transmission and thereby minimize channel switching overheads such as switching delay, energy consumption, and the associated data loss. Among the published multichannel MAC protocols for WSNs, only a few consider channel quality before assigning wireless channels. Whereas to the best of our knowledge, only few contemplates both channel quality and stability during stream-based communication, however, not very suitable for varying (noisy) environment. To bridge this gap, this work proposes a novel dynamic Multichannel Adaptive approach for Grading Immediate Channels (MAGIC). The MAGIC protocol may predict and classify the available channels under channel quality and stability assessment methodology and determines Local Preferred Channels (LPCs) in a noisy environment. Afterwards, the best among those channels may be selected (under some QoS criteria) for stream-based communication in WSNs. The proposed MAGIC algorithm minimizes channel processing overheads by considering only non-blacklisted channels during a communication session. Simulation results show that the MAGIC algorithm achieves better performance than the counterparts in terms of channel switching delay, energy consumption, and the associated throughput loss.