No single sensitive clinical outcome measure is capable of reliably reflecting the broad spectrum of clinical manifestations seen in multiple sclerosis (MS). Disease activity in groups of patients with MS is more readily detected by magnetic resonance imaging (MRI) of the brain and spinal cord than by neurological examination. However, the appearance of new lesions and progression of total lesion area as detected by MRI fails to predict reliably sustained progression of neurological impairment in individual patients. For this reason MRI is accepted as a primary outcome measure in exploratory clinical trials, while sustained progression of neurological impairment is the preferred primary outcome measure for pivotal phase II/III trials that lead to the approval of new treatments for MS.A comparison of proportions of patients meeting treatment failure as defined by discrete data at a specified time has previously been used in clinical trials of treatments for MS. Fewer patients may be needed to detect treatment effects when treatment failure is defined by continuous data that are analysed in a time-to-failure framework. Also, it may be possible to have nurses or trained technicians, rather than clinicians, administer tests that are continuous measures of neurological function.It is anticipated that recommendations provided by the Task Force on Clinical Outcomes for MS sponsored by the National Multiple Sclerosis Society of the US will include new multidimensional disease-specific composite outcome measures that better reflect the broad range of clinical expression and activity of MS. However, it is unclear whether investigators are prepared to abandon neurologist-based ratings as the principal method for determining disease severity and response to treatment. More sensitive clinical outcomes may improve the predictive validity of certain MRI techniques and facilitate expeditious testing of promising therapies in trials involving fewer patients.