The aim of this study was to obtain reliable data about the current aetiology (i.e. the frequency of the individual pathogens) of community-acquired pneumonia (CAP) while surveying the diagnostic and therapeutic behaviour of Italian chest physicians, compared with existing guidelines, and to test the usefulness of the current severity "criteria" or score as a predictor of disease outcome and guide for appropriate hospitalization. A prospective multicentre observational trial was carried out between October 1994 and February 1996 by the Italian Association of Hospital Pneumologists (AIPO) study group on respiratory infections. A total of 613 consecutive patients suffering from CAP were enrolled in 25 centres throughout Italy. Clinical, radiological and microbiological data were collected and patients were followed-up until complete resolution or death. Aetiological tests were not carried out in 204 patients. In the remaining 409 cases, the aetiology was defined by serological and quantitative microbiological tests in 184 (44.9%) patients. A total of 194 strains of pathogen were detected. The most frequently detected micro-organism was Streptococcus pneumoniae (18.5% of pathogen strains) but, unlike in other series of patients, high percentages of intracellular pathogens (32.5%, all with serological confirmation, mostly due to Chlamydia pneumoniae (13.4%) and of Gram-negative enterobacteria and Pseudomonas aeruginosa (12.5%) were also found. Antibiotic treatment differed from that recommended in American Thoracic Society guidelines, with a greater use of third-generation cephalosporins. Overall, a higher rate of hospitalization and a lower death rate than in other comparable studies was observed.