INTRODUCTION: Recent advances in high-density microelectrode arrays (MEAs) have allowed clinicians to record from individual neurons (single-unit activity) or from ensembles of neurons (multi-unit activity, MUA) in the human brain. These recordings have yielded fundamental advances both in epilepsy and in normal brain function. However, the properties of bursts of neuronal activity during seizures have not been characterized. METHODS: Five patients at our institution were implanted with microelectrode arrays and met inclusion criteria. Bursts were captured using custom code (MATLAB). We used statistical techniques to compute the rate similarity or sequence similarity between any two sequences. RESULTS: We found strong internal consistency within groups of pathological discharges -- that is, IED and seizure related bursts -- for both rate and sequence similarity. We found that baseline, IED, and seizure-related bursts all reside in particular subspaces, after dimensionality reduction using UMAP. During seizures, the LFP exhibits strong wavelike properties. Interestingly, the MUA direction echoes that of the LFP in this patient. However, MUA discharges are only partially wavelike. In IEDs, while the LFP is quite wavelike, the MUA does not exhibit wavelike properties. We found that LFP wave directions, for both seizures and IEDs, tend to arise from the seizure source, as determined using the ECoG. The relationship between the ECoG source and the MUA wave direction is less consistent. Seizure sequences start near the baseline subspace. Over time, seizure bursts are recruited to a distinct subspace, at which time they develop wavelike properties. CONCLUSIONS: Seizure events induce wavelike states in neuronal bursts. These findings may help explain why seizures alter cognition.