Many patients with presumed nosocomial pneumonia probably have infiltrates on the chest radiograph, fever, and leukocytosis resulting from noninfectious causes. Because of the high mortality and morbidity associated with nosocomial pneumonias, however, most clinicians treat such patients with a 2-week empiric trial of antibiotics. Before therapy is initiated, the clinician should rule out other causes of pulmonary infiltrates, fever, and leukocytosis that mimic a nosocomial pneumonia (e.g., pre-existing interstitial lung disease, primary or metastatic lung carcinomas, pulmonary emboli, pulmonary drug reactions, pulmonary hemorrhage, collagen vascular disease affecting the lungs, or congestive heart failure). If these disorders can be eliminated from diagnostic consideration, a 2-week trial of empiric monotherapy is indicated. The clinician should treat cases of presumed nosocomial pneumonia as if P. aeruginosa were the pathogen. Although P. aeruginosa is not the most common cause of nosocomial pneumonia, it is the most virulent pulmonary pathogen associated with nosocomial pneumonia. Coverage directed against P. aeruginosa is effective against all other aerobic gram-negative bacillary pathogens causing hospital-acquired pneumonia. The clinician should select an antibiotic for empiric monotherapy that is highly effective against P. aeruginosa, has a good side-effect profile, has a low resistance potential, and is relatively inexpensive in terms of its cost to the institution. The preferred agents for empiric monotherapy for nosocomial pneumonia are cefepime, meropenem, and piperacillin. Single organisms are responsible for nosocomial pneumonia, not multiple pathogens. S. aureus rarely, if ever, causes nosocomial pneumonia but is mentioned frequently in studies based on cultures of respiratory tract secretions. S. aureus, unless accompanied by a necrotizing pneumonia with rapid cavitation within 72 hours, in the sputum indicates colonization rather than infection and should not be addressed therapeutically. Antibiotics associated with a high resistance potential should not be used as monotherapy or included in combination therapy regimens (i.e., ceftazidime, ciprofloxacin, imipenem, or gentamicin). Combination therapy is more expensive than monotherapy and is indicated only when P. aeruginosa is extremely likely, based on its characteristic clinical presentation, or is proved by tissue biopsy. Therapy should not be based on respiratory secretion cultures regardless of technique. Optimal combination regimens include cefepime or meropenem plus levofloxacin or piperacillin or aztreonam or amikacin. Nosocomial pneumonias usually are treated for 14 days. Lack of radiographic or clinical response to appropriate empiric nosocomial pneumonia monotherapy after 14 days suggests an alternate diagnosis. In these patients, a tissue biopsy specimen should be obtained to determine the cause of the persistence of pulmonary infiltrates unresponsive to appropriate antimicrobial therapy.