1. The haemodynamic effects of different narcotic agents have been tested in healthy rats and in rats with cirrhosis. 2. Pentobarbital suppresses the sympathetic nervous system. Susceptibility to ketamine is unpredictable, leading to both insufficient pain relief and narcosis related mortality. The combination diazepam-fluanisone induces profound hypotension. After insertion of catheters, awake, freely moving rats are stable and not distressed. This allows repeated measurements after manipulation. Moreover, procedure-related mortality is low and rats have a better stress response. 3. In the awake animal, arterial pressure is 126 +/- 10 for healthy animals, and 111 +/- 16 and 102 +/- 10 mmHg for cirrhotic animals without and with ascites, respectively (P = 0.018). The respective values for portal pressure are 6.9 +/- 1.4, 11.6 +/- 2.5 and 16.2 +/- 2.9 mmHg (P = 0.0001). After a bleeding, arterial pressure is better preserved than portal pressure in the three groups (P < 0.0001). Plasma volume in cirrhotic rats exceeds that of healthy rats. Plasma renin activity, aldosterone and catecholamines do not differ between the groups studied. In cirrhotic rats with and without ascites, glomerular filtration rate tends to be higher (P = 0.12), renal plasma flow is elevated (P = 0.001) and filtration fraction is lower (P = 0.002) than in healthy rats. 4. In conclusion, haemodynamic experiments in the cirrhotic rat should be performed in the awake rat. Arterial hypotension, impaired filtration fraction, enlarged plasma volume and portal hypertension are present in cirrhosis before the development of ascites. This can as well be explained by splanchnic pooling of blood, as by peripheral vasodilatation. The decrease in portal pressure with preserved arterial pressure after a bleeding protects cirrhotic rats from ongoing variceal bleeding.