Cleaning is an essential step in the reprocessing of GI endoscopes. Because manual cleaning is tedious and technician dependent, automated cleaning may have some advantages over the manual method. In simulated use testing, the efficacy of automated cleaning (System 83-2+DT, Custom Ultrasonics) was compared with a standardized manual cleaning method previously reported by the investigators (Amer J Gastroenterol 1999;94:1546). The automated reprocessor simulates the mechanical effects of a brush or sponge by pulsing enzymatic solution through most channels at a frequency of 60 Hz, and against the exterior surface of the insertion tube at 40,000 Hz. METHODS: One Olympus CF100TL videocolonoscope was contaminated with a simulated soil consisting of 2 parts sterile milk powder, 10 parts fetal calf serum, 10 parts Mycobacteria chelonae, for a final mycobacterial concentration of 10 9 CFU/mL. The suction housing, air-water housing, valves, insertion tube, instrument channel and auxillary channel were independently inoculated, and then air-dried for 60 min prior to cleaning. The size of inoculum (total CFU) was established for each site through control experiments. Each inoculated site was flushed with sterile phosphate buffer solution; the effluent was poured through a 0.45μm membrane filter, cultured in Middlebrook 7H11 agar, and incubated for 10 days at 37° C. Three to five experiments were conducted for controls and for manual and automated cleaning methods using the same scope. RESULTS: See table. CONCLUSIONS: (1) In simulated use testing under worst-case scenario, automated cleaning achieved a 3 to 5 log reduction in mycobacterial bioburden, which was at least equivalent to standardized manual cleaning. (2) Manual cleaning of the auxillary channel is potentially problematic because its small channel caliber precludes the use of a cleaning brush. (3) Used in conjunction with rigorous manual cleaning, automated cleaning may add an extra margin of safety for reprocessing of flexible GI endoscopes.