When driving in fatigue-sensitive zones of urban tunnels, fatigue onset is challenging to detect, alertness levels decrease with fatigue aggravation, and the continuous fatigued state becomes increasingly challenging to counteract. To address these issues, three distinct visual-intervention schemes are developed based on tunnel-sidewall rhythms: a repetitive rhythm for fatigue reminders, a staggered combined rhythm for fatigue warnings, and an undulating combined rhythm for fatigue arousal. Next, a simulated driving platform is constructed, incorporating a blank set of no-rhythmic modes and the aforementioned tunnelsidewall rhythms. Drivers′ visual behaviors (fixation and saccade) across different lane positions (left, middle, and right) are analyzed to quantitatively assess their visual-fatigue risks, and the results indicate the following. (1) Rhythm patterns significantly influence visual-fatigue risk, with the observed risk intensity following the order: control group > repetitive rhythm > staggered combined rhythm > undulating combined rhythm, confirming that sidewall-rhythm incorporation and optimization effectively mitigate visual fatigue. (2) Fatigue-sensitive zones exhibit distinct spatial-distribution characteristics across lanes, with visual fatigue varying significantly by lane position in the following order: left lane > right lane > middle lane. (3) The left lane emerges as the most prominent 〖HJ2mm〗fatigue-sensitive zone because of its narrowest effective field of view and immediate proximity to the tunnel sidewall, which increase visual stress. (4) Based on the effectiveness of rhythm patterns, lane-sensitive characteristics, and engineering feasibility, the following crucial optimization methods are proposed: the staggered combined rhythm should be prioritized in fatigue-sensitive zones of urban tunnels, whereas the most effective undulating combined rhythm should be prioritized where conditions permit.