The Federal Motor Vehicle Safety Standards currently require driver-side rearview mirrors to be flat. For rearview mirrors of typical size, this requirement normally results in a blind zone on the driver side that is large enough to conceal an average size passenger car. In recent years a number of studies have suggested that nonplanar rearview mirrors may be an effective solution to this problem. This paper reviews the evidence on possible effectiveness of nonplanar mirrors, assesses the strength of that evidence, and makes tentative recommendations. The main conclusion is that the use of nonplanar mirrors would probably result in a net gain in safety, but that the effectiveness of the mirrors is likely to depend on details of how they are implemented. Issues that should be resolved by additional research (some of which is already underway) are: (1) How would U.S. drivers respond to a mixed fleet of vehicles, some of which had flat mirrors and some of which had nonplanar mirrors? (2) If nonplanar mirrors are allowed, what range of designs (from among spherical convex mirrors of various radius of curvature, and different versions of aspheric mirrors) should be used?