Our understanding of motor-related processes has been significantly influenced by the description of the event-related desynchronization and synchronization phenomena in the mu (~8-13 Hz) and beta (~13-30 Hz) frequency bands [1,2,3]. Accordingly, non-invasive brain-computer interface (BCI) applications exploiting motor imagery (MI) usually depend on spatially- and band-limited power changes as the brain markers of interest. Yet, numerous studies have questioned the role of signal power as the best descriptor of movement-related brain activity modulations. On a single-trial level, beta band activity is characterized by short, transient, and heterogeneous events termed bursts rather than sustained oscillations [4,5,6].In a recent study we analyzed the activity of channels C3 and C4 from multiple, open EEG datasets [7] with participants performing MI tasks. We demonstrated that a beta burst analysis of hand MI binary classification tasks (i.e. “left” vs “right”) is often superior to beta power in terms of classification score [8].Here we expand upon this approach. We introduce an algorithm that constructs convolution kernels based on beta burst waveforms and creates classification features that are comparable to those of state-of-the-art methods. By independently transforming the EEG recordings with each kernel we compute a proxy of the waveform-resolved burst rate that is then spatially filtered using the common spatial pattern algorithm (CSP) [9]. Then, all spatial features are combined and used for classification. We adopt a time-resolved decoding strategy and show that, when compared to CSP-based signal power in the beta and mu bands, the waveform-resolved burst rate often yields superior decoding results only needing a short amount of data points, likelybecause it captures slow (mu-like) signal changes.This analysis shows that BCI applications could benefit from utilizing such features and paves the way for a real-time implementation of the proposed methodology.