Transcranial alternating current stimulation (tACS) is increasingly used as a tool to non-invasively modulate brain oscillations in a frequency specific manner. A growing body of neuroscience research utilizes tACS to probe causal relationships between neuronal oscillations and cognitive processes or explore its capability of restoring dysfunctional brain oscillations implicated in various neurological and psychiatric disease. However, the underlying mechanisms of action are yet poorly understood. Due to a massive electromagnetic artifact, overlapping with the frequency of interest, direct insights to effects during stimulation from electrophysiological signals (i.e. EEG/MEG) are methodologically challenging. In the current review, we provide an overview of analysis approaches to recover brain signals in M/EEG during tACS, detailing their underlying concepts as well as limitations and methodological and interpretational pitfalls. While different analysis strategies can achieve strong attenuation of the tACS artifact in M/EEG signals, a compete removal of it is not feasible so far. However, we argue that with a combination of careful experimental designs, robust outcome measures and appropriate control analyses, valid and important insights to online effects of tACS can be revealed, enriching our understanding of its basic underlying mechanisms.