The timely, accurate recognition of adverse drug reactions (ADRs) presents a fascinating and important challenge for the health care system. At the premarketing level, there must be a delicate balance in requiring clinical studies that are of adequate power to define the benefits and risks of a medication but that are not so cumbersome or lengthy as to drag on excessively, preventing the introduction of useful new agents into practice. The premarketing phase is complicated further by the fact that it is statistically impossible for large numbers of important high-risk patients to be included to answer every plausible question about adverse effects. In the case of the elderly, improvement still needs to be made in including sufficient numbers of typical "old" elderly, in addition to the 66-year-old all-stars sometimes presented as adequate geriatric representation. For some frail or demented elderly, as for pregnant women and children, ethical concerns may arise