In a multi-laboratory trial, minerals-modified glutamate medium (MMGM) was compared with lauryl tryptose lactose broth (LTLB) in the multiple tube method for the enumeration of coliform organisms, including Escherichia coli, in water. Samples of raw and chlorinated waters yielded a total of 2313 positive tube-reactions with MMGM and 2174 with LTLB. These were interpreted either as E. coli; other coliform organisms; or as false positive reactions. The results at first reading (18 or 24 h) and at 48 h have been analysed statistically in terms of (i) most probable numbers of coliform organisms; (ii) positive reactions and their interpretation; and (iii) whether or not the sample yielded any E. coli or other coliform organisms. All three analyses indicated the same trends. For the detection of E. coli in raw waters LTLB was better than MMGM at 18–24 h, but MMGM was better at 48 h with waters containing small numbers of coliform organisms; for raw waters with greater numbers of organisms, both media performed equally well. Analysis of a subset of samples read at both 18 and 24 h indicated that the superiority of LTLB over MMGM with raw waters disappeared by 24 h. For chlorinated waters, LTLB yielded more positive gas reactions at 18–24 h, but fewer of these were E. coli than with MMGM; at 48 h MMGM was clearly better than LTLB for total coliform organisms including E. coli – especially if the numbers were small. MMGM therefore remains the medium of choice for the detection of E. coli as an indicator of faecal contamination of chlorinated drinking water supplies. It is also better for the detection of small numbers of E. coli in other waters.