The activity state of neuronal networks can be characterized by the spatial-temporal grouping of their action potentials given a sufficiently large simultaneous recording sample. A sequence of action potentials (spike train) often has high frequency spike episodes that are generally called bursts. However, bursts are difficult to quantify and require operational definitions that reflect the type of activity and the interest of the experimenter. This paper presents a simple method for defining bursts as strings of spikes with only two parameters: a minimum number of spikes per burst and a maximum interspike interval. These two values represent a simple parameterization that is adequate for the description of temporal grouping in spike trains. Because this method has a minimal computation time, it allows implementation of burst analysis in real-time, including statistical changes in burst variables, histograms of burst types, and patterns in combinations of burst variables.