Disinfected and non-disinfected samples have been used to determine the accuracy of the ISO procedure (ISO 9308-1) for detection of E. coli in drinking water. Samples were analysed using the ISO procedure at both 36 and 44 degrees C and using the defined substrate technology method Colilert-18/Quanti-Tray (Colilert-18). Utilizing the confirmation procedure described in ISO 9308-1, large numbers of false positive E. coli results were obtained using the ISO primary isolation procedure at 36 degrees C. However, when glucuronidase production was used as the confirmation procedure, the 'confirmed' count of E. coli was lowest with ISO 9308-1 performed at 36 degrees C. When TTC medium was incubated at 36 degrees C confirmation using production of indole at 44 degrees C resulted in 29% more 'E. coli' being recovered than when confirmation was performed using production of glucuronidase. When 44 degrees C was used as the primary isolation temperature the difference between the number of 'confirmed' E. coli identified using the two confirmation procedures was less than 1% and was not significant. Identification of isolates which 'confirmed' when using production of indole at 44 degrees C as the test method but which failed to produce beta-D-glucuronidase, showed that the majority of these isolates were Klebsiella oxytoca.