Pain ratings from 268 migraine patients have been used to compare the visual analogue scale (VAS) and a four-point verbal rating scale. All patients completed pain ratings on both scales at the beginning of a migraine attack and 4 h after starting treatment with sumatriptan or placebo. The VAS scores showed large variability within each category on the verbal rating scale. A common way of analysing pain in migraine studies is to classify patient response as a success or failure based on the score on the four-point verbal rating scale. In this study, the statistical power of analysing response either as “success or failure” or by means of the VAS score has been investigated by stochastic simulation. The simulations showed that the two response measures resulted in approximately equal power.