Within 15 min of endotracheal Intubation, the I8SOlutlon of pulmonary edema wesstudled over the next 12 h In 34 mechanically ventllatad patients by (1) serial measurements of the alveolararterial oxygen difference,(2) the extent of edema on the Initial and follow-up chest radiograph, and (3) by an Initial and final measurement of total protein and albumin concentretlon In sequential samples of pulmonary edama fluid. Based on the oxygenation and chest radiographic data, 24 patients clinically Improved and 10 patients did not Improve. In the 10 patients who did not clinically improve (3, hydrostatic edema; 7, permeability edema), there was no change In the final edema fluid protein concentration (4.1±1.1 gl100 ml) compared with the initial edema fluid protein concentration (4.2±1.0 gl100 ml)(p= ns). However, In the 24 patients who clinically Improved (15, hydrostatic edema; 9, permeability edema), there was an Increase In every patient's final edema protein concentration (5.6±2.3 g/100 ml) compared with their Initial edema protein concentration (3.8±1.2 gl100 ml)(p< 0.01). In 13 of these 24 patients, the final edema fluid concentration (7.3

±1.6 g/100 ml) exceeded the final plasma protaln concentration (5.6±0.8 gl100 ml) by a mean velue of 1.7 g/100 ml protein. The data provide the first evidence In humans to support the hypothasis that active Ion transport across the alveolar epithelial barrier is the primary mechanism for clearance of edema fluid from the air spaces of the lung. In addition, the use of sequential pulmonary edema fluid protein concentration as an Index of alveolar epithelial barrier function provided a good prognostic Indicator of survival In the 16 patients with Increased permeability pulmonary edema.