Relatively simple objective criteria are now available to predict which patients are at risk for bad outcomes from community-acquired pneumonia. In general, these include older patients and those with certain coexisting illnesses (especially neoplastic disease) or findings of altered mental status, hypotension, severe tachycardia, tachypnea, fever, acidemia, azotemia, hypoxemia, hyperglycemia, anemia, or hyponatremia. The major causes of severe pneumonia are S pneumoniae, H influenzae, and L pneumophila. Less common causes include mixed aerobic and anaerobic mouth flora, as well as M pneumoniae, C pneumoniae, gram-negative bacilli, and S aureus. Specific diagnosis is hampered by a lack of reliable diagnostic tests, but Gram's stain of expectorated sputum and cultures of sputum and blood may occasionally be helpful. Many empirical treatment regimens have been recommended, including those of the American Thoracic Society and the Infectious Diseases Society of America, which are reviewed here. It is hoped that better diagnostic tools will permit future targeting of microbes with narrow-spectrum therapy to diminish the risk of selection of resistant strains with empirical regimens.