Water samples of various origins were inoculated into a specific coliform-selective lactose broth provided with lipoic (thioctic) acid, and the time evolution of the redox potential of the cultures was monitored during incubation at 41 degrees C by use of gold versus reference electrodes. Positive potential-time responses, i.e., 100-mV potential shifts recorded within 20 h of inoculation, were related to the initial number of fecal coliforms in the broth determined by control enumeration techniques, and the organisms responsible were isolated and identified by conventional procedures. A total of 30 samples of wastewater, 38 of surface water, 553 of groundwater, and 110 of drinking water were tested successively. A total of 240 natural water samples, including 172 groundwater samples, and 1 drinking water sample were found to be positive in the potentiometric test. The majority (i.e., 92.5%) of the relevant potentiometric detection times were shorter than 15 h, and 96% of these could be attributed to Escherichia coli. Fifteen hours corresponded to the limit for detecting 1 E. coli cell per 100 ml of water. About 78% of the potentiometric responses occurring after 15 h were induced by fecal coliforms other than E. coli (Enterobacter cloacae, Klebsiella pneumoniae, and Citrobacter freundii). Calibration curves relating detection times shorter than 15 h to fecal coliform (i.e., E. coli) concentrations were constructed for the natural water samples tested. There were minor variations in the average growth rate of the organisms in the relation to the contamination level of the water tested. The number of false-positive samples in the potentiometric test was equivalent to that of false-negative samples (groundwater or drinking water).