Monitoring describes the prospective supervision, observation, and testing of an ongoing process. The result of monitoring provides reassurance that the goal has been or will be achieved, or suggests changes that will allow it to be achieved. In therapeutics, most thought has been given to Therapeutic Drug Monitoring, that is, monitoring of drug concentrations to achieve benefit or avoid harm, or both. Patients and their clinicians can also monitor the progress of a disease, and adjust treatment accordingly, for example, to achieve optimum glycaemic control. Very little consideration has been given to the development of effective schemes for monitoring for the occurrence of adverse effects, such as biochemical or haematological disturbance. Significant harm may go undetected in controlled clinical trials. Even where harm is detected, published details of trials are usually insufficient to allow a practical monitoring scheme to be introduced. The result is that information available to prescribers, such as the Summary of Product Characteristics, frequently provides advice that is incomplete or impossible to follow. We discuss here the elements of logical schemes for monitoring for adverse drug reactions, and the possible contributions that computerized decision support can make. We should require evidence that if a monitoring scheme is proposed, it can be put into practice, will prove effective, and is affordable.