As new therapies become available for the treatment of multiple sclerosis, the relative value of established and newer disease-modifying therapies must be considered. However, comparing the apparent efficacy of different agents across clinical trials is not easy and can be misleading when different therapies have been studied during different time periods. There has been a shift in current clinical trials toward enrolling patients with less advanced or less active disease compared with trials undertaken when no effective therapies were available. If early treatment is more effective than late treatment, this practice will produce a bias in favor of newer agents. Head-to-head trials offer the most reliable means of comparing therapies, but these trials are expensive and time consuming. Consequently, cross-trial comparisons are necessary, but a reliable means to make such comparisons is needed. One useful (but imperfect) approach is to compute the relative risk of therapy and the number-needed-to-treat, applying both measures to any cross-trial comparison. These measures capture different aspects of the trials (relative and absolute differences) and, if they agree, this suggests that the cross-trial comparison may be valid. If the two methods disagree, no reliable conclusion about relative efficacy can be made. There are only two valid conclusions from the available head-to-head and cross-trial data. First, high-dose interferon-β (IFNβ)-1a or IFNβ-1b subcutaneous has a greater impact than weekly IFNβ-1a IM on several clinical and MRI outcomes. Second, high-dose IFNβ-1a or IFNβ-1b subcutaneous has a similar clinical impact to glatiramer acetate, although IFNβ subcutaneous is superior on some MRI outcome measures.