Published Phase III immunomodulatory treatment trials in relapsing and remitting multiple sclerosis have demonstrated a modest decline in attack rates, but only a minor effect on disability. As genuine disability progression is difficult to ascertain in relatively short studies with the conventional rating scales available, the acquisition and analysis of relapse data are critical. However, there are as yet unresolved questions related to the latter. We will first discuss the problems associated with relapse definitions by trial investigators, the paucity of the data collected (especially on the magnitude and duration of exacerbations) and statistical issues in their analysis. We will then suggest practical points for obtaining more accurate information on relapses and evaluating them meaningfully. While there is still general consensus among neurologists that primary endpoints for therapeutic trials should be clinical, improvements for future protocols are essential.