In a prospective study to evaluate the diagnostic yield of different microbiological tests in hospitalised patients with community-acquired pneumonia, material for microbiological investigation was obtained from 262 patients. Clinical samples consisted of the following: sputum for Gram staining, culture, and detection of pneumococcal antigen; blood for culture and serological tests; urine for detection of Legionella pneumophila serogroup 1 antigen and pneumococcal antigen; and specimens obtained by fiberoptic bronchoscopy. A pathogen was identified in 158 (60%) patients, with Streptococcus pneumoniae (n=97) being the most common causative agent of community-acquired pneumonia. In 82% of the 44 patients with an adequate sputum specimen, a positive Gram stain was confirmed by positive sputum culture. S. pneumoniae infections were detected principally when adequate sputum specimens were examined by Gram stain and culture and when adequate and inadequate sputum specimens were tested for the presence of pneumococcal antigen (n=58; 60%). The urinary pneumococcal antigen test was the most valuable single test for detection of S. pneumoniae infections (n=52; 54%) when sputum pneumococcal antigen determination was not performed. Fiberoptic bronchoscopy was of additive diagnostic value in 49% of the patients who did not expectorate sputum and in 52% of those in whom treatment failed. Investigation of sputum by a combination of Gram stain, culture, and detection of pneumococcal antigen was the most useful means of establishing an aetiological diagnosis of community-acquired pneumonia, followed by testing of urine for pneumococcal antigen. Fiberoptic bronchoscopy may be of additional value when treatment failure occurs.