So far, the MPN technique has been the method chosen for detecting pathogenic bacteria in waters, since it allows us to use larger volumes. In the evaluation, however, comparatively frequently (10 … 15%) nontypical codes occur. The validity of the Poisson model is confirmed for pure cultures by extensive laboratory experiments with pure cultures and controlled mixed cultures. In mixed cultures, however, a comparatively large number of atypical codes occur, since obviously a single organism is not sufficient for the formation of a growth‐cover on the test tube, so that the true number of germs may be considerably underestimated. Obviously, the accompanying flora has an inhibiting effect on the salmonella detect