Subject reports of efficacy of treatment of haemophilia-related joint bleeding are by definition subjective, yet are often the primary outcome in studies comparing therapies. Verbal descriptors such as effective, partially effective, poorly effective, not effective are treated as dichotomous or categorical variables in analyses, lowering the statistical power relative to that which might be achieved with a continuous variable. The aims of this study were to examine reports of pain recorded on a 100-mm visual analogue scale (VAS) during the course of joint bleeding; determine whether pain varied by treatment period among pairs reporting discordant outcomes on a verbal scale (one product effective, the other not effective); test whether the two products under study were equivalent with respect to VAS scores; and evaluate their relationship to verbal reports of efficacy. Data from the international, prospective, randomized, crossover FEIBA NovoSeven Comparative study of two bypassing agents used for treatment of 96 bleeding episodes in 48 participants were examined. VAS scores were associated with verbal descriptors of efficacy at every time point, and were equivalent between treatment periods. There were differences in mean scores at time points at which participants rated one treatment effective, the other not effective. As a continuous variable, the VAS score may have more power than a dichotomous variable and when used with verbal descriptions of efficacy can improve the overall accuracy of assessment. This report highlights an important consideration in the selection of outcome measurement that can be generalized to other haemophilia treatment research.