Wildlife habitat is fragmented by transportation systems, a legacy of land-use and a continuing, often un-mitigated impact of transportation. The primary tools to reduce these impacts are discretionary projects to construct “wildlife crossing structures” (WCS) and associated fencing. Although WCS are relied upon to mitigate impacts to wildlife, little attention is paid to animal responses to noise and artificial light at night in designing these structures and the approaches to them. We describe an approach that considers wildlife responses to light and noise from traffic to design and test model approach zones to WCS. We used commonly-available landscape design software to create virtual approaches to the proposed Liberty Canyon Over-Crossing (LCOC) across US 101 in Southern California. When completed, LCOC will be the largest and possibly most-costly WCS in the world. We created berm and wall structures in the northern approach to the LCOC, which is particularly exposed to traffic disturbance. We used previous research about noise and light impacts on wildlife, as well as field measurements of maximum and average noise (A-weighted decibels) and light (illuminance) to inform placement and dimensions of the structures. For the berms, we used residual fill on-site to create the structures, with a target of keeping the total volume of the berms less than the available fill. For the walls running alongside the highway and across the LCOC, we used a combination of conventional solid walls and walls backed by an earthen slope. All structures were created to be feasible, meaning slopes of 1:1 to 2:1, wall heights less than 20 feet tall, and attention paid to drainage. We virtually tested the light and noise attenuation for 4 variants – 1) WCS only, 2) WCS + walls, 3) WCS + walls + one large berm, and 4) WCS + walls + 3 staggered berms to permit better water flows. Light conditions were tested for a virtual animal approaching the LCOC from north or south where relative headlight brightness in the animal’s view was estimated using image processing software. Noise conditions for each model were tested using the FHWA Traffic Noise Model where thresholds of 55 dBA and 65 dBA were used as important performance criteria for “semi-natural” and “tolerable” conditions for the target wildlife species. Variant (3) performed the best in terms of light (no headlight glare) and noise attenuation (<55 dBA), because it provided a continuous barrier to noise and light through a large part of the approach zone. Although the several berm model would allow water to move between the berms, noise and light was able to penetrate through the gaps for water, reducing the effectiveness. We propose that this approach could be used for all future WCS where approach zones and the structures themselves may be impacted by traffic noise and light.