According to Sherrington (1910), hind limb withdrawal to noxious stimulation of the skin is considered to be a monolithic flexion reflex such that wherever the stimulus is applied the same response is observed, namely flexion at the hip, knee and ankle. Contrary to this view, an accumulation of evidence suggests that extensors are also involved in the withdrawal response with the actual muscles recruited being dependent on the location of the noxious stimulus (Clarke et al. 1989; Schouenborg & Kalliomaki, 1990). The conclusions of these studies have been based on findings using electrophysiological recording in hind limb muscles rather than measurement of actual movements, therefore the aim of the present study was to use an image analysis system to address this issue. Experiments were performed on 14 rabbits lightly anaesthetized by a continuous i.v. infusion of Saffan (between 4 and 30 mg kg^-1 h^-1). Reflex responses were evoked by mechanical pinch of the skin using toothed forceps fitted with a pressure transducer (mean force 1.4 ± 0.1 kg for 1.6 ± 0.2 s). Stimuli were applied to eight locations on the hind limb but only data from pinching the left heel or toes are reported here. Reflex movements of the left hind limb were measured using a CODAmotion movement analysis system which enabled angles at the hip, knee and ankle joints to be calculated via detection of strategically located infra-red emitters. The first response to mechanical pinch at the heel was extension of the ankle at a latency of 245 ± 44 ms (mean ± S.E.M.). This was followed by significantly later flexion at the knee and hip (348 ± 51 and 352 ± 49 ms, respectively; repeated measures ANOVA, p = 0.01). The mean angle change for extension at the ankle was 6 ± 1.5 deg and values for flexion at the knee and hip were 12 ± 1.1 and 9.8 ± 1.3 deg, respectively. In contrast, there was no significant difference in the latency of the ankle, knee and hip responses to noxious mechanical stimulation of the toes (559 ± 100, 530 ± 89 and 561 ± 89 ms, respectively; repeated measures ANOVA, p = 0.4) but these values were significantly longer compared to heel pinch (paired t tests, p 0.05). By measuring actual movements, these data show that extension of the ipsilateral hind limb is an important part of the withdrawal response to noxious stimulation. Furthermore, the contribution of extension to reflex withdrawal is dependent upon the site of the stimulus and the joint studied.