The process by which new therapies enter clinical practice is frequently suboptimal. Often, ideas for new therapies are generated by clinical observations or laboratory studies; therapies based on those ideas may enter clinical practice without any further scrutiny. As a consequence, some ineffective practices become widespread. This article proposes a six-stage protocol for the implementation of new therapies. Hypotheses about therapy based on preclinical research should be subject to clinical exploration and pilot studies prior to rigorous assessment with randomised clinical trials. If randomised clinical trials suggest that the intervention produces clinically important effects, further randomised studies can be conducted to refine the intervention. New interventions should not be recommended, or included in teaching curricula, or taught in continuing education courses until their effectiveness has been demonstrated in high-quality randomised clinical trials.