The visual analogue scale (VAS) is a standard measurement tool in pain research and clinical practice, and has been shown to have linear scale properties for mild to moderate pain. Our aim was to evaluate the scaling properties of the VAS in subjects with severe acute pain. After Ethics Committee approval we studied 22 patients and asked them to rate the severity of their pain on a 100 mm VAS at the initial assessment (VAS1), and again after administration of analgesic medication. The subject was asked to nominate when they considered their pain intensity had halved, and at this time they were asked to rate this on a second VAS (VAS0.5). When the subject had received satisfactory relief of their pain, they were asked to describe how much their pain had been relieved and were then asked to rate their final pain state using a third VAS (VASfinal). The mean (SD) scores were VAS1 84 (14) (range 56-100), VAS0.5 42 (13) and VASfinal 21 (16). The mean (95% CI) for VASratio was 0.51 (0.45-0.57). The mean (SD) patients' estimate of pain relief was 77 (21)% from that of baseline, with a mean (SD) VASfinal 0.26 (0.20), 95% CI 0.17-0.38. The correlation of the patients' estimate of pain relief with the VASfinal was r=0.89, rho=0.87, both P<0.001. The VAS is a linear scale in subjects with severe acute pain. Changes in the VAS score represent a relative change in the magnitude of pain sensation.