Background: The non invasive diagnosis of amebic liver abscess allows the use of empirical therapy without the requirement of invasive diagnostic procedures. Aim: To determine the discriminatory capacity of clinical, laboratory and ultrasound studies for the etiological diagnosis of liver abscess. Patients and methods: Sixty one patients were initially included in this prospective study, but 12 did not comply with the inclusion criteria. Of the rest, 29 (59%) had an amebic liver abscess, 16 (33%) had a pyogenic liver abscess and four (8%) had an abscess of mixed etiology. Blood cultures were done in 42 patients. Ultrasound guided needle aspiration was done in 7 patients with amebic liver abscess and 13 patients with non amebic liver abscess. Results: The clinical picture and ultrasound fndings were similar in all types of amebic abscess. ELISA test for IgG anti-Entamoeba histolytica antibodies were positive in 100% of patients with amebic liver abscess. Antibodies measured by gel diffusion were positive in 93%. All patients with mixed liver abscess had positive antibodies and some of them positive culture. Blood cultures were positive for anaerobic bacteria in five patients. Cultures of aspirated material were positive in 7 patients (obligate anaerobic bacteria in 3 and facultative anaerobic bacteria in the rest). The most common complications, whatever the etiology, were right pleural effusion and systemic inflammatory response. Conclusions: A final model of binomial regression analysis revealed that age under 40 years, an hematocrit greater than 35% and an elevation in prothrombin time of less than 1.5 seconds had enough discriminatory capacity for the diagnosis of amoebic liver abscess (Rev Méd Chile 2003; 131: 1411-20).