Medicines derived from plants, such as digoxin, morphine, and vincristine, are important in conventional Western therapy. These examples also demonstrate that natural is not synonymous with innocuous, since these medicines have narrow safety margins. However, as with other conventional medicines, their licensing is based on three criteria: efficacy in a given indication; acceptable safety in usual therapeutic use; and quality of manufacture. Should we judge herbal medicines by the same criteria?

There are some herbal medicines of demonstrable efficacy: for example, in one trial a standard extract of Hypericum (St John's wort) was as effective as paroxetine in depression.1 For most herbal treatments, however, good trials of efficacy are lacking, and conducting them would be expensive. Ernst noted two years ago that systematic reviews provided good evidence of efficacy for just 11 herbal medicines and had found “promising but not convincing” results for nine more.2 Herbalists since the Englishman Nicholas Culpeper in the 16th century have held forth promises, but most …