A prospective standardized collection of clinical, microbiological and pharmaceutical information on antibiotic use was conducted in Australia and New Zealand intensive care units (ICUs) involving 481 consecutive critically ill patients who were receiving antibiotics for any reason while in ICU. Patients had a mean SAPS II score of 34.1 +/- 17.8 with an expected mortality of 15.6% (actual mortality 12%). Of these, 292 (60.8%) were admitted to the ICU within 72 hours of surgery. Among such surgical patients, 233 (79.9%) received antibiotics for "surgical prophylaxis" while in ICU (48% of sample population). The second largest group of patients treated with antibiotics in ICU included those with systemic inflammatory response syndrome and clinical suspicion of infection (38%). Antibiotics were prescribed for the treatment of clinically diagnosed infection in 268 patients. Clinical response was apparent in 62.6% and in most (71%) was achieved in the first 72 hours of treatment. The incidence of antimicrobial-related side-effects was 4%, mostly in the form of diarrhoea or rash (75% of all side-effects). The most commonly prescribed antimicrobials were gentamicin (n = 146), ceftriaxone (n = 98), vancomycin (n = 94) and metronidazole (n = 111). Three times daily prescription of aminoglycosides was uncommon (< 1%). Forty-one patients had a documented infection (positive culture) with a gram-negative organism. Of these, 17 received therapy with a single antibiotic and 24 received therapy with two antibiotics. Despite similar illness severity, there were six deaths in the former group and only two in the latter.