During the subcooled blowdown phase of a large-break loss-of-coolant accident (LB-LOCA), if the break occurs in the cold leg, coolant flows backward, forming a vertically downward two-phase flow. Void fraction is a key parameter in two-phase flow, but research on its measurement in vertically downward two-phase flows within rod bundle channels remains limited. This study established an experimental setup for downward two-phase flow. The test section, made of acrylic, had a square flow channel (50 mm × 50 mm) containing a 3 × 3 rod bundle to simulate fuel assemblies. The differential pressure method and a wire-mesh sensor (WMS) were used to measure void fraction under room temperature and atmospheric pressure. WMS data validated the differential pressure method results. The effects of gas–liquid flow ratio, spacer grids, and channel height on void fraction were analyzed. Results show that gas–liquid flow ratio and channel height significantly influence void fraction: as superficial gas velocity increases, void fraction rises. The rod bundle and spacer grids induce flow pattern changes, causing unstable void fraction variations over time. Due to buoyancy effects, void fraction at mid-channel height exceeds that at both ends. Comparisons with WMS measurements confirm the differential pressure method’s reliability for void fraction measurement in rod bundle channels, with errors below 8%. Additionally, high-speed camera observations revealed five flow patterns, and their formation mechanisms were analyzed.