Community-acquired pneumonia (CAP) is an infectious illness that frequently motivates hospital admission when comorbid conditions are present. However, the epidemiology of CAP in relation to the underlying disease of the patients is not well known. We performed a prospective multicenter study with the aim of assessing the clinical characteristics, etiology, and outcome of chronic obstructive pulmonary disease (COPD) patients with CAP. Between October 1992 and December 1994 we studied 124 COPD patients (mean FEV1 40 +/- 11% of predicted, mean FVC/FEV1 49 +/- 10) admitted because of CAP to one of the participating centers. An attempt to obtain an etiologic diagnosis was performed by means of blood cultures (n = 123), sputum cultures (n = 97), pleural fluid cultures (n = 17), protected specimen brush samples (n = 41), percutaneous transthoracic needle aspiration (n = 41), and serology (n = 106). Etiologic diagnosis was achieved in 80 (64%) of cases, however, diagnosis based upon valid techniques was only possible in 73 (59%) cases. The main causal microorganisms were the following: Streptococcus pneumoniae in 32 (43%), Chlamydia pneumoniae in 9 (12%), Hemophilus influenzae in 7 (9%), Legionella pneumophila in 7 (9%), Streptococcus viridans in 3 (4%), Coxiella burnetii in 3 (4%), Mycoplasma pneumoniae in 2 (3%), Nocordia asteroides 2, Aspergillus ssp. 1, and others 10. In three of these cases the etiology was polymicrobial. Bacteremia was present in 19 (15%) cases; S. pneumoniae was the most frequent isolate (13 cases). Antibiotic treatment was modified in 22 cases due to etiologic findings, and in 9 due to therapeutic failure. Ten patients died (8%), and 22 needed mechanical ventilation, the mortality rate in the latter population being 23%. Total or partial resistance of S. pneumoniae to penicillin was observed in 10 of 32 (31%) isolations, and to erythromycin in 2 (6%). The results of this study are important for the standardization of empiric antibiotic strategies in COPD patients with pneumonia.