To study the effect of enterally administered polymyxin E, tobramycin, and amphotericin B (selective flora suppression) on bacterial colonization, infection, resistance, and mortality rate. Prospective, consecutive crossover controlled study. Two surgical ICUs in a university hospital; ICU I with ten beds, ICU II with eight beds. Two hundred patients entered the 1-yr trial. Fifty of 111 patients received selective flora suppression during the first 6 months in ICU I (test group), while 61 of 111 patients served as the control group in the following 6 months. In ICU II, 49 of 89 patients received no selective flora suppression in the first 6 months (control group), followed by 40 of 89 patients receiving selective flora suppression during the second 6-month period (test group). The test group got a mixture of nonabsorbable antibiotics (paste and suspension) in the digestive tract. The control group received paste and suspension without antimicrobial agents. All 200 patients received cefotaxime during the first 4 days. With the use of selective flora suppression, colonization with aerobic Gram-negative bacilli was significantly (p less than .01) reduced. There was also a significant reduction in nosocomial bronchopulmonary (ICU I and II; p less than .001) and urinary tract (ICU II; p less than .001) infections. The difference in mortality was not significant. There was no development of resistance against the antibiotics used during the limited period evaluated. Selective flora suppression is effective in reducing secondary colonization by aerobic Gram-negative bacilli. Reduction of bronchopulmonary and urinary tract infections most likely occurs with colonization prevention.