Questions have been raised about the correctness of water quality models with complete mixing assumptions in cross junctions of water distribution systems. Recent developments in the mixing phenomenon within cross junctions of water distribution networks (WDNs) have heightened the need for evaluating the existing incomplete mixing models under real-world conditions. Therefore, in this study, two cross junctions with pipe diameters of 100 × 100 × 100 × 100 mm and 150 × 150 × 150 × 150 mm were employed in laboratory experiments to evaluate six existing incomplete mixing models for 25 flow rate scenarios ranging between 1.5 and 3.0 l/s. It was observed that within the same flow rate scenario, the degree of mixing in a cross junction with a pipe relative roughness of 6.00 × 10−5 (pipe diameter of 25 mm) was higher than that in a cross junction with a pipe relative roughness of 3.00 × 10−5 (pipe diameter of 50 mm) and smaller. Considering the real-world size of pipes in evaluating the incomplete mixing models showed that two incomplete mixing models, AZRED and the one by Shao et al., had the best accordance with the results of the laboratory experiments.