Abstract: To address safety risks associated with driving at tunnel entrances in foggy conditions, a comprehensive evaluation model was developed by integrating data from both driving behavior and physiological indicators. Through road tests, driving behavior parameters (longitudinal acceleration, steering wheel angle, braking frequency, etc.) were collected from 10 drivers under four levels of fog visibility (L1: 0–50 m to L4: 500–1,000 m). while simultaneously recording physiological indicators such as pupil diameter and heart rate variability, to reveal the dynamic mechanisms by which fog visibility affects driving safety. The results indicate that reduced visibility significantly alters driving patterns, reflecting a coupled characteristic of increased visual-psychological load. Safety thresholds were defined using the interquartile range method. A combined weighting approach was employed to integrate subjective weights from the Analytic Hierarchy Process (AHP) with objective weights from the entropy weighting method. Risk levels were quantified through fuzzy comprehensive evaluation, revealing that the “unsafe” membership degree at tunnel entrances and exits under Level 1 visibility reached 12.6%, and 12.9% at the tunnel entrances and exits under L1 visibility, respectively—an increase of nearly fourfold compared to L4 visibility. Visibility and risk exhibit a significant negative correlation.