We tested the hypothesis that cerebral malaria is caused by blood-brain barrier inflammation and cerebral edema. In a group of 157 Thai patients with strictly defined cerebral malaria, cerebrospinal fluid (CSF) opening pressures were normal in 79% and were lower in fatal cases than in survivors (means +/- 1 SD, 144 +/- 58 and 167 +/- 51 mm CSF, respectively, P = 0.051). CSF: serum albumin ratios (X 10(3)) in 39 of them were significantly higher than in 61 British controls (medians 8.5 and 5.5, respectively, P = 0.04), but were no higher in 7 fatal cases. In a group of 12 patients this ratio was not significantly higher during coma than after full recovery (means +/- 1 SD, 9.0 +/- 6.2 and 6.7 +/- 4.2, respectively, P greater than 0.1). CSF alpha 2-macroglobulin concentrations were always normal. CSF : serum 77Br- ratios were elevated in 11/19 comatose cases but fell to normal 4 to 9 days later in 11/11 cases. Dexamethasone treatment had no significant effect on bromide partition. The percentage of an intravenously administered dose of 125I-human serum albumin detectable per ml of CSF 6 hr after intravenous injection was 2.4 +/- 1.3 X 10(-5) in 14 comatose patients and 4.4 +/- 4.0 X 10(-5) in 9 of them during convalescence (P greater than 0.1). These results demonstrate that the blood-CSF barrier is essentially intact in patients with cerebral malaria and give no support to the idea that cerebral edema is the cause of coma.