Currently, regulatory agencies have serious difficulty in obtaining high-quality scientific proof that demonstrates the clinical efficacy of new medical devices. These difficulties are specific to medical devices and to the medical environment that uses them. Schematically the clinical development of a new medical device has two stages: feasibility studies and studies to demonstrate clinical benefits. Feasibility studies are proposed immediately after the preclinical phase. The type of study that is methodologically appropriate is a noncomparative trial that responds to questions about patient selection, the development of implantation techniques, clinical efficacy, and complications. The demonstration of clinical benefits depends on the performance of a randomized control trial, although the feasibility studies are taken into account. The construction of the trial should be based on the formulation of a clear, specific, and pertinent principal objective. Eligible patients should correspond to those for whom the new device is intended in daily practice. The choice of a control group depends on the reference strategy or treatment, determined from the literature. A single principal endpoint should be proposed, consistent with the principal objective, which should be clinical (whenever possible), pertinent, and validated. The measure used to determine the endpoint must be as objective as possible. Multicenter trials are preferable to facilitate patient recruitment and minimize the inclusion period. Moreover, the results of multicenter studies can be extrapolated more readily. Nonetheless, the teams likely to participate in a multicenter trial must have stabilized their learning curve. To meet the methodological requirements of clinical trials for new medical devices, clinical research must improve its structure, especially by promoting the links between industry, clinicians, and academics.