To determine the influence of circulatory factors on onset of neuromuscular block, we have measured twitch height in an arm with a tourniquet inflated during onset and compared this with data from a control arm in 20 patients under fentanyl-thiopentone-nitrous oxide-isoflurane anaesthesia. Patients were allocated randomly to receive either vecuronium 0.1 mg kg-1 (n = 10) or suxamethonium 1 mg kg-1 (n = 10). The EMG response of the first dorsal interosseous to single twitch stimulation of the ulnar nerve every 10 s was recorded in both arms. When neuromuscular block was 20% (i.e. twitch height was 80% of control), the tourniquet was inflated to a pressure of 250 mm Hg. It was deflated 5 min later. In the vecuronium group, the rate of onset did not differ in both arms and mean maximum block was 95 (SD 4)% in the tourniquet arm, which was not different from 99 (2)% in the perfused arm. In the suxamethonium group, the presence of a tourniquet decreased the rate of onset by 66%. Maximum block was only 74 (20)% in the tourniquet arm compared with 97 (5)% in the perfused arm (P < 0.05). The difference in maximum neuromuscular block between arms was 4 (3)% in the vecuronium group and 22 (17)% in the suxamethonium group (P < 0.01). We conclude that during onset, neuromuscular block continued to increase in spite of interruption of blood flow, and this increase was greater with vecuronium than with suxamethonium.(ABSTRACT TRUNCATED AT 250 WORDS)