The ability to differentiate cardiac from permeability edema on the basis of clinical and radiographic criteria was studied in 70 ICU patients in whom subsequent pulmonary artery catheterization (PAC) was performed. Our study demonstrated that the clinical assessment of permeability pulmonary edema was correct in 17 of 20 patients (85%). In contrast, of the 50 patients initially suspected of having cardiac edema, only 31 (62%) were predicted correctly (p less than 0.05). Complications relating to catheterization occurred in 25% of patients, with 3 deaths. We conclude that the diagnosis of cardiogenic pulmonary edema, based on clinical criteria alone, is often inaccurate in the intensive care setting. The failure of patients to respond to initial therapy should mandate pulmonary artery catheterization, despite the attendant risks. Furthermore, even though the clinical diagnosis is correct in 85% of patients with permeability pulmonary edema, PAC data may be necessary for optimal management.