Neural oscillations have been associated with different functions and disorders of the brain. Recently, the non-sinusoidal waveform of oscillations has gained interest as a biomarker for neurological and psychiatric diseases, and as an indicator of underlying network physiology. However, waveform analysis in the time domain has been limited to dominant neuronal oscillations and by pre-selection of waveform features. Here, we contribute a novel Fourier-series waveform analysis that is noise resistant and provides a complete description of waveforms. We applied Fourier-based waveform analysis to human cortical oscillations recorded using magnetoencephalography (MEG). This approach allowed us to robustly dissociate multiple spectrally and spatially overlapping cortical rhythms across theta, alpha and beta frequency ranges in the human brain. Fourier-based waveform analysis is a powerful new tool to identify and distinguish neuronal rhythms, and to study their spectral fingerprints in health and disease.