Perhaps no frequent driving maneuver is more hazardous than the left turn. Existing statistical analyses indicate that the older drivers are over represented in the left turn configuration. It is not surprising that the left-turn proves such a hazardous configuration since the turning driver has minimal, obscured, and conflicting information upon which to base their turn decisions. In addition, understanding the problems of the left-turn presents a number of information and task decomposition challenges. For example, given a driver's expected vehicle response, prediction of other vehicles future positions relative to the driver's own position must be visually interpreted from available motion-in-depth information. Some have likened the task of the turning driver to one of coincident timing, where turn initiation and completion must be synchronized with acceptable gaps in on-coming traffic (2). Acceptability is predicated on each individual drivers perception of this traffic in terms of physical characteristics (3, 4). Objectively, vehicles can vary in terms of their relative approach velocities, the changing gap difference between themselves and the vehicle they follow, and their configuration in terms of size, shape, and color. If drivers use these physical variables singly, changes in turn strategies would logically be consistent with such physical parameters. If, however, drivers base their decisions on higher-order information sources like rate-of-expansion of the vehicle frontal surface (time-to-arrival), the pattern of results would not be consistent with manipulations of these physical properties. Previously, this proposition was tested for a college-age population in the University of Minnesota's fixed-based automobile simulator (5). This group of drivers, with a mean age of 24.2, initiated left-turns, not on the basis of any physical metric, but through inference on time-to-arrival information. The present experiment examines the same driving maneuver in older drivers. We hypothesized that the turn strategies employed by older drivers would be, in part, mediated by these same higher-order information sources, but that the scaling of that information relative to their own self-perceived limitations would render them more conservative in a manner consistent with traditional performance speed assessment metrics such as visual search time and reaction time.