It is common clinical practice to estimate the degree of neuromuscular blockade by tactile evaluation of twitch responses. The aim of the present study was to evaluate the use of tactile responses of adductor pollicis to double-burst stimulation (DBS) and train-of-four (TOF) peripheral nerve stimulation for monitoring moderate and profound levels of neuromuscular blockade. The study comprised 44 women scheduled for gynaecological laparotomy and anaesthetised with midazolam, fentanyl, thiopentone, halothane, nitrous oxide and atracurium. The tactile responses of the adductor pollicis were compared with mechanomyographical measurements in the contra lateral arm during recovery from neuromuscular blockade. The observers (anaesthetic nurses) of the tactile responses were blinded with regard to the stimulation pattern and the mechanomyographical measurements. The time from injection of the initial dose of atracurium until tactile reappearance of the first twitch in DBS (D1), was 24.6 (0-39.8) min, median (range). This was more rapid than the time until tactile reappearance of the first twitch in TOF (T1) 32.8 (18.2-43.4) min (P < 0.05). The median time from tactile reappearance of D1 until T1 recovered to 15% of the control twitch height was longer than the median time from tactile reappearance of T1 (14.6 versus 10.5 min) (P < 0.05). One or two responses to DBS or TOF were often felt before any responses had been detected mechanomyographically in the contralateral arm. When three or four responses to TOF were felt, responses were always detected mechanomyographically. It is concluded that tactile evaluation of responses ot DBS stimulation can estimate deeper levels of blockade than tactile evaluation of responses to TOF.