Cerebral blood flow determinations using the Xenon inhalation test in two patients with transient global amnesia showed overall decreases during the attack with a more marked fall in temporal-inferior frontal flow in one case and in temporal flow in the other. In the latter patient the temporal anomaly was reversible after naloxone injection but clinical improvement was lacking. Blood flow determinations in a further patient with transient global amnesia, after 2 mg of ergotamine tartrate plus caffeine, showed a global decline in flow without focal anomaly, with recovery after ten days. Values obtained in two other patients 24 and 38 hours respectively after typical transient global amnesia attacks were normal. The effect of naloxone appears related to a direct neuronal action although the role of endorphins in this affection is not conclusive. Comparison of these anomalies with those noted by Reichle with position emission tomography imaging suggests a primary neuronal mechanism.