No extant protocol for an experimental trial of therapy in multiple sclerosis provides a reliable tool for the determination of therapeutic efficacy. Quantitative evaluations of clinical phenomena do not reflect the true activity of the underlying disease, and summative scoring of neurologic dysfunction is unreliable. Comparisons in treatment trials should utilize data only from subjects with substantive changes or no changes. Evaluation of experimental treatment of acute exacerbations is unreliable and less important than that aimed at eradication or prevention. Knowledge of etiology and pathogenesis remains the best potential source of effective therapy. Suggestions are made for conducting trials of empiric methods along simpler lines with high powers of resolution. Serious reservations should be retained concerning the wisdom of further expenditure of time, energy, or funds for experimental trials of therapy in multiple sclerosis patterned after protocols utilized to date.