Transmural variceal pressure was measured by fine-needle puncture following a 250-micrograms bolus injection of somatostatin given intravenously, together with a slow infusion of 250 micrograms/h, and compared with results following a 500-micrograms bolus and with placebo in 19 patients with cirrhosis who had recently bled. Pressure was recorded continuously for a 4-min period. Placebo did not alter variceal pressure (n = 6), whereas a 250- or 500-micrograms bolus produced a rapid decrease in pressure from 33 +/- 7 cmH2O to 20 +/- 6 (n = 7) and from 35 +/- 5 cmH2O to 24 +/- 6 (n = 6), respectively. The difference between the two groups was not significant. This decrease occurred within 30 to 90 s, but was followed by a progressive increase in variceal pressure over the next 2 min in all except two patients. After 3-4 min, variceal pressure still remained below pretreatment levels. Further studies are needed to evaluate the optimal infusion rate. These results seem to indicate that future clinical studies using somatostatin in the treatment of variceal haemorrhage should include repeated bolus injections to arrest variceal bleeding.