While in animal experiments neurogenically initiated pulmonary edema is a well known event and is supposed to be due to centrally initiated hemodynamic disturbances ("neurohemodynamics") in patients with severe cerebral lesions fulminant alveolar edema is reported to occur very rarely. The questions addressed by this study are: 1. whether and to what extent changes in extravascular lung water (EVTVL) can be demonstrated in patients with a severe isolated cerebral lesion; 2. whether a relationship between the severity of the cerebral lesion and accompanying EVTVL changes can be proven; and 3. whether or not EVTVL changes are associated with corresponding changes in intravascular hydrostatic and oncotic Starling parameters; i.e. cardiogenic or noncardiogenic pulmonary edema accompanying the cerebral lesion. This study included 44 patients presenting with a severe isolated cerebral lesion and decerebrate posturing on admission. EVTVL (by thermo-dye double-indicator technique), pulmonary gas exchange (AaDO2/pAO2), colloid oncotic pressure (COP) and mean systemic arterial (SAP), mean pulmonary arterial (PAP), and pulmonary capillary wedge pressures (PCWP) were measured from the day of admission to the 6th day after the acute cerebral lesion maximally; in addition the microvascular pressure in the pulmonary bed and intravascular filtration pressure were calculated from the above mentioned parameters. The neurological status on admission and throughout the observation period was scored using the Innsbruck Coma Scale (ICS) and the neurological outcome by the Glasgow Outcome Scale (GOS). Statistical analysis was performed using the distribution independent Kruskal Wallis test, the correlation coefficient r (Pearsan and Bravais), and the Spearman rank correlation (RSp); values are given as means +/- SEM; the significance has been set at P less than 0.05. Our results reveal an overall increase in EVTVL from 8.8 +/- 0.8 ml/kg on the day of admission up to 11.3 +/- 1.6 ml/kg on the 4th day. While survivors (n = 13) remained within the normal range of EVTVL (less than 9 ml/kg), non-survivors (n = 31) started at an already elevated level (10.05 +/- 1.04 ml/kg) and reached their maximum values (15.4 +/- 2.3 ml/kg) on day 3 to 4. In 3 non-survivors these increased initial EVTVL values were accompanied by pathologically increased intravascular pressures, indicating that hydrostatic mechanisms were involved in the EVTVL rises. While the hydrostatic pressures normalized spontaneously, EVTVL values stayed within the pathological range throughout the remaining observation period.(ABSTRACT TRUNCATED AT 400 WORDS)