The J-turn, also known as the restricted crossing U-turn and Superstreet, is an innovative geometric design that can improve intersection safety. Although this design has been in use in several states for many years, there is very little research-based guidance for several design parameters. A driving simulator study was conducted to analyze the parameters of lane configuration, U-turn spacing, and signage. Two lane configurations were examined: ( a) an acceleration–deceleration configuration, in which acceleration and deceleration lanes are provided, and ( b) a deceleration-only configuration, in which only deceleration lanes are provided. Lane configuration was found to be the most important parameter affecting J-turn safety according to speed differentials. The only significant interaction effect among parameters was between lane configuration and U-turn spacing. The acceleration–deceleration configuration performed better than the deceleration-only configuration, having 66.3% fewer safety critical events. Vehicle trajectories and average lane change locations showed that U-turn spacing significantly affected the acceleration–deceleration configuration (e.g., average merge locations changed by 96% to .0101%) but not the deceleration-only configuration. No strong preference was demonstrated for either the directional or the diagrammatic signage style. This paper presents one of the first human-factors studies of the J-turn to focus on developing design guidance. This human-factors approach complements traditional approaches such as crash analysis and microsimulation.