SUMMARYβ‐Adrenoceptor blockers always change splanchnic haemodynamics in cirrhotic patients. Azygous blood flow, as a measure of collateral circulation including that through varices, is always reduced, but the effects on portal pressure, whether measured directly or by the wedged hepatic venous pressure, are variable. The initial correlations between a 25% reduction of resting pulse rate and similar percentage reduction in the wedge‐free hepatic venous gradient, has not been reproduced in subsequent studies. Therefore, to study the effect of changes in haemodynamic indices and the likelihood of variceal bleeding, direct measurements of such indices need to be made in clinical trials. At present there are no haemodynamic or clinical factors which can be used to select patients who will have a good therapeutic response to propranolol other than those documented in the first clinical trial of propranolol for the prevention of variceal re‐bleeding from Paris. Thus the hypothesis that β‐adrenoceptor blockers may lessen the incidence of bleeding in cirrhotics, by partially reducing portal pressure or flow or both, needs testing in further clinical studies. The selection criteria of the first clinical trial of propranolol in Paris need to be confirmed. Two subsequent trials, in which patients were not selected but in which many patients had similar clinical characteristics to the Paris patients, could not confirm a therapeutic effect of propranolol. No fatal complications due to propranolol administration have been reported in cirrhotic patients. Complications are reversible. Pharmacological treatment including β‐adrenoceptor blockade appears ideal for trials of primary prevention of variceal bleeding. Some preliminary results including use in decompensated cirrhotics are encouraging. However, as for trials for prevention of re‐bleeding, the design and analysis of such trials needs careful evaluation to take into account the outcome of patients who discontinue medication, whether due to simple non‐compliance or due to side‐effects, and also the influence of abstinence from alcohol on ble