In 1990 there was a sudden increase in the incidence of colonization and infection due to Acinetobacter baumannii (AB) in our intensive care units (ICUs). The isolates were multiply resistant to beta-lactam and aminoglycoside antibiotics, but remained susceptible to imipenem, amikacin, and ampicillin-sulbactam. We examined the frequency of infection and colonization with AB and the effects of increased imipenem and amikacin therapy on Pseudomonas aeruginosa. We also used disease-matched controls to determine the clinical and financial impacts of treating colonization. All patients with at least one AB isolate from January-December 1992 were identified retrospectively and classified as infected or colonized based on published Centers for Disease Control criteria; the control group was selected from a computerized medical records data base matching primary diagnostic codes (102 patients both groups). The 102 patients yielded 140 isolates, 124 resistant AB and 16 sensitive AB. Thirty three patients were infected, 69 colonized. Mortality correlated with APACHE II scores. Patients acquired the organism approximately 2 weeks after admission; they had a mean ICU stay of 27.35 days, compared with 5.53 days for controls. Patients with positive AB cultures required significantly more use of ventilators than those with negative AB cultures. They also had significantly longer hospital stay, more bed transfers, greater duration and number of antibiotics, and higher hospital and pharmacy charges. Unnecessary treatment for colonization with either imipenem or amikacin resulted in a substantial decrease of P. aeruginosa susceptibility to each agent. The financial impact of treating colonization was significant and is a potential area for cost avoidance. Our results emphasize the need to extubate and move patients to non-ICU beds as soon as possible to decrease the risk of nosocomial infection. It also highlights the need to avoid treating colonization, thus avoiding unnecessary antibiotic therapy.