Although many studies in neuroscience are based on comparing neuronal responses to single, isolated sensory or motor events, multiple events frequently occur in close temporal proximity in freely moving animals. This often obscures the precise temporal correlation between each event and the relevant brain activity. By simulating neuronal responses in multi-event tasks, we show that perievent time histograms (PETHs) greatly distort the underlying true responses. We propose a multi-event deconvolution method that can separate the contribution of each event to the overall neuronal activity. The improvements over PETH in analyzing real data are demonstrated using simulated data and a sample electrophysiological recording obtained from rats in a task involving responses to a reward predictive cue.