OBJECTIVE: To compare the efficacy of endoscope disinfection using automated and manual systems. DESIGN: Prospective randomized trial. SETTING: A 1,000-bed tertiary care referral center. METHODS: All endoscopes underwent a three-stage decontamination process including brushing and cleaning with water and detergent, manual or automated disinfection with 2% glutaraldehyde, and 70% alcohol rinse with forced air drying. Cultures were obtained from endoscopes from both groups before and after alcohol rinse and then after overnight storage. RESULTS: Cultures from 8/30 (27%) automated and 11/30 (37%) manually disinfected (P= 0.58) endoscopes grew gram-negative bacteria and/or nontuberculous mycobacteria before the alcohol rinse. After alcohol rinse, 3 (10%) of 30 automated and 8 (27%) of 30 manually disinfected endoscopes remained contaminated (P= 0.28). Manually disinfected endoscopes were contaminated more frequently with coliform bacteria, whereas endoscopes undergoing automated disinfection were more frequently contaminated with nontuberculous mycobacteria, but the differences were not statistically significant. After alcohol rinse and forced air drying, there was no difference in contamination rates between freshly disinfected endoscopes and those stored overnight (7/30 (23%) versus 4/30 (13%), P= 0.50). Colonoscopes and duodenoscopes were contaminated more often than gastroscopes (P= 0.00001). CONCLUSION: The persistent endoscope contamination after manual and automated disinfection indicates the importance of developing more reliable and effective disinfection methods. (Infect Control Hosp Epidemiol 1993;14:383-389).