Blocking effects and wall interference represent longstanding, unresolved challenges in the field. In bridge wind tunnel experiments, a large test model is often preferred to accurately capture the detailed structural features of the model's surface. However, excessively large test models can lead to significant blockage or wall interference effects, which may distort the experimental results. Changes in model size also introduce Reynolds number effects, which are coupled with the blockage effect, thereby complicating the analysis. In this study, computational fluid dynamics (CFD) simulations and physical wind tunnel experiments were employed to analyze a segment model of a streamlined closed box girder. The wind speed and model size were independently varied in the simulations to investigate the effects of Reynolds number and blockage. Additionally, physical experiments were conducted in different wind tunnels to validate the CFD simulation results. This paper first identifies that under typical test conditions, the Reynolds number effect is much smaller than the blockage effect, allowing the decoupling of the two phenomena to isolate their respective influences. Further results demonstrate that increasing model size and angle of attack leads to significant changes in the aerodynamic force coefficients. The larger the model size and the greater the angle of attack, the faster the rate of change in these coefficients. Moreover, due to the model's asymmetry, behaviors slightly differ under positive and negative angles of attack. Based on CFD simulations and experimental results, this paper proposes recommended critical blockage ratios for typical test conditions, specifically within an angle of attack range of −3° to 3°: 3.9% for the lift coefficient, 4.1% for the drag coefficient, and 4.3% for the moment coefficient, which are all smaller than the prevalently adopted value of 5%. Empirical formula incorporating an exponential correction is also provided.