Computerized polysomnographic systems have came into common use in sleep laboratories around the world. Despite potential advantages over standard paper polysomnography, these computerized systems have been minimally evaluated as to accuracy, analysis time, or cost effectiveness when compared to paper. We evaluated the Healthdyne ALICE 3 system for comparability to paper polysomnography in sleep quantification and technician analysis time. Fifty patients were recorded simultaneously both on paper and on the ALICE 3 system and analyzed blindly with summary data from these records being quantified and compared. Five additional patients were studied for epoch-by-epoch analysis. Score-rescore assessments were accomplished for both groups. The results indicate that when allowed to autoscore, this computerized system produced substantial errors in sleep staging (REM sleep time 56.4 + 4.9 minutes vs 73.2 + 8.4 minutes for paper versus computer). This was the case for respiratory (AHI of 26.5 + 4.3 vs 15.3 + 2.6 for paper vs computer) and arousal assessment as well. However, with editing, similar results to those obtained with paper were achieved (REM sleep time -56.4 + 4.9 vs 59.0 + 4.6; AHI -26.5 + 4.3 vs 26.1 + 4.7 for paper and computer respectively), with differences rarely exceeding score-rescore discrepancies. Analysis time was substantially reduced by use of the computer (172.6 + 9.9 vs 79.7 + 4.8 minutes for paper vs computer). Epoch-by-epoch analysis revealed a trend to score toward wakefulness or lighter sleep on computer compared to paper although the differences were small. Respiratory, arousal and PLM scoring were quite similar. In conclusion, this study suggests that the ALICE 3 system with editing can produce results similar to those obtained with paper.