To determine the etiology of community-acquired pneumonia in patients treated in an ambulatory setting, using serological methods, and to compare presenting symptoms, radiographic manifestations, and clinical outcomes of patients with pneumonia of "atypical" and undetermined etiology. This prospective cohort study was conducted in emergency room and outpatient facilities of Victoria General Hospital, Halifax, Nova Scotia, and in offices of participating family doctors based in Halifax. One hundred forty-nine adults with acute onset of one or more symptoms or signs suggestive of pneumonia and radiographic evidence of pneumonia who provided informed consent were enrolled. Patients known to be HIV positive or who had been discharged from a hospital within the previous 10 days were ineligible for enrollment. Demographic features and clinical data were collected by direct patient interview and chart review by trained research nurses. Outcome measures included quantitative evaluation of pneumonia-specific symptoms, and responses to the Short Form 36 Health Survey at presentation and at 30 days after presentation. Information was also collected on each patient's health prior to pneumonia, as well as the time until each patient's self-reported return to work and to usual activities. The etiology of pneumonia was determined by testing acute and convalescent serum samples for antibodies to Legionella pneumophila serogroup 1, Mycoplasma pneumoniae, Chlamydia pneumoniae, Chlamydia psittaci, Coxiella burnetii, adenovirus, respiratory syncytia virus, influenza viruses A and B, and parainfluenza viruses 1, 2, 3. The study population consisted of 149 patients, 54 (36%) of whom were men, with a mean age (+/- SD) of 41 +/- 15 years. An etiological diagnosis was made in 74 (49.7%) patients using serological methods. Etiological agents included M pneumoniae 34 (22.8%); C pneumoniae 16 (10.7%); M pneumoniae and C pneumoniae 5 (3.4%); C burnetii 4 (2.7%); influenza A virus 4 (2.7%); and other agents 6% (7.4%). Three patients (2%) had a conventional bacterial etiology, and 72 patients (48.3%) had pneumonia of undetermined etiology. Patients with pneumonia of known (atypical) and undetermined etiology were similar in terms of age, gender, race, education, employment, and comorbidity. Despite a higher proportion of patients with pneumonia of known etiology reporting sweats, chills, and headache at presentation, the two groups were similar for symptom severity and bother. The patients with pneumonia of undetermined etiology were more likely to have multilobar pneumonia (P < 0.02). Both patients with atypical pneumonia and those with pneumonia of undetermined etiology suffered severe deterioration of physical functioning with a marked but incomplete recovery at 30 days. Those with atypical pneumonia had higher physical functioning and general mental health scores at 30 days. Nearly half the cases of ambulatory community-acquired pneumonia are due to "atypical" agents. It is not possible to reliably distinguish patients with atypical pneumonia from those with pneumonia of undetermined etiology by clinical features at baseline. The outcomes in terms of resolution of symptoms, functional status, return to work, and return to usual activities are essentially similar in the two groups.