In recent years a number of studies have attempted to rank drugs by a single measure of harmfulness as the basis for decisions about scheduling and classification. These efforts are fundamentally flawed, both conceptually and methodologically. The effort to provide a single measure masks the variety of non-comparable dimensions that are relevant, the fact that benefits are ignored for most, but not all, drugs and that the harms of a drug are not invariant to the policy regime chosen. Methodologically, the most prominent recent effort ignores drug interactions and mixes aggregate and individual harms inappropriately. Instead we suggest that multiple dimensions of harm need to be displayed to inform human judgments of what drugs should be scheduled. Harm is not usefully reducible to a single dimension, and even perfect rankings would not constitute a ‘sufficient statistic’ for determining scheduling decisions.