An automated system that emulates the visual analysis of sleep records (electroencephalograms, electrooculograms, and electromyograms) by depicting relevant waveforms (features) and evaluating the sleep stage according to the standard scoring manual is presented. Algorithms for detection of K-complexes, spindles, rapid eye movements, electromyographic bursts, delta, alpha, and mixed frequency waveforms based on morphological parameters that reflect visual recognition, perform in the 91 to 98% correct range. The staging algorithms that were developed use logic structures that emulate human decision-making. The somnogram obtained on the first full-night (eight-hour) sleep record that was analyzed is in 92% agreement with the somnogram provided by an experienced sleep clinician. The automated system detects all the features online and completes the staging in 5 min offline.