A two-year analysis was performed at our hospital to evaluate the incidence and clinical impact of bloodstream infections (BSI) in medical, surgical, and intensive care unit patients. During 1999-2000 there were 521 cases of BSI with an incidence of 10.1/1,000 admissions. The most frequent predisposing factors for BSI were intravascular catheter (56.4%) and previous use of antibiotics (50.9%). Infections were considered as primary in 67.9% of cases. Urinary tract and intravascular catheter were the most frequent source of secondary bacteremia (43.1% and 35.9%, respectively). At the time of the first positive blood culture, 83.5% of patients were receiving empirical treatment, but it was adequate in only 53.9% of cases. After antimicrobial susceptibility testing, adequate antibiotic treatment was given in 67.9% of cases. Statistical analysis of independent risk factors showed that mortality was significantly related to age (P < 0.048), rapidly-fatal diseases (P < 0.001), septic shock (P < 0.020), multiorgan failure (P < 0.001), previous use of antibiotics (P < 0.008), Enterobacteriaceae producing extended-spectrum beta-lactamases (P < 0.036), and inadequate empirical treatment (P < 0.039). Based on local epidemiology and susceptibility data, microbiology laboratories should periodically release recommendations on the optimal empirical treatment for different wards.