Interlayer voids threaten pavement integrity, but their ground-penetrating radar (GPR) signatures and measurable dimensions depend on the acquisition conditions and target geometry. This study combines two-dimensional finite-difference time-domain simulations, a full-scale wet-sand experiment, and a preliminary field case study. Nine numerical configurations represent air- or water-filled voids. Physical targets comprise polystyrene blocks and water-bag-filled polycarbonate boxes surveyed using stepped-frequency 3D-GPR. Under the adopted processing and display convention, the two physical target groups show contrasting polarity patterns, and larger targets generally have clearer lateral boundaries. Quantitative analysis shows that dimensional accuracy is configuration-specific and does not support a universal size threshold. Under the investigated acquisition and interpretation conditions, thickness cannot be reliably estimated for either the air-like or water-rich targets. Field coring and endoscopy, supplemented by the reported excavation measurements, verify one void with a plan extent of approximately 5.2 m × 1.2 m and a depth range of 0.44–0.54 m. This case supports preliminary correspondence between controlled and field signatures, rather than general field-accuracy benchmarks.