The high rate of septicaemias (20%, 19% and 14%) observed in our Intensive Care Unit (ICU) during the first 3 years was due to an epidemic incidence of Serratia sp. (S) (26% during the first year) and Klebsiella sp. (K) (25% during the third) and decreased significantly in the following 6 years (mean incidence of 11%) (p less than 0.01). During this epidemic phase these organisms were isolated quite frequently (between a 14% and a 6%) from all patients admitted. The K was more regularly present, for the mean time intervals free of its bacteriological presence were shorter (11 days) than those of S (27 days) (p less than 0.01). The K was isolated in more patients (160) than S (79) (p less than 0.01) and in more samples (360) than S (235) (p less than 0.01), but caused less secondary septicaemias per colonized patient (7% versus 29%) (p less than 0.01). In 59% of all S septicaemias the organism was previously isolated in other culture, while this was observed in only 34% of K septicaemias (x2 = 3.78, p = 0.052). The large variations in the incidence of septicaemias within our ICU, the appearance of sequential epidemic outbreaks, with a different behaviour of S and K and the individual risk of septicaemia of patients colonized by these organisms are noted.