Computer methods can automatically recognize interictal spikes and seizures quite reliably, but they still make a large number of false detections because of physiological or artifactual nonepileptiform transients. In the context of long-term monitoring, these methods can be used efficiently and safely despite their imperfection. They allow considerable data reduction, by selecting epileptiform activity and discarding the largest part of the recording. In addition to data reduction, computer methods can be useful in refining the analysis of epileptic seizures, revealing information not available from visual inspection of the paper tracing. Recent technological advances make these applications affordable in a clinical EEG laboratory.