A constant cross‐section wind tunnel usually experiences a negative pressure gradient (ie, favorable pressure gradient, FPG), resulting in a certain degree of discrepancy between the simulated and idealized atmospheric boundary layer (ABL) flows. To investigate its impact, three large‐eddy simulation (LES) cases were conducted in this study: one simply reproduced a wind tunnel, one removed the FPG by adjusting the ceiling height, and one reduced the FPG by canceling unnecessary boundary layers associated with the tunnel sidewalls and ceiling. The required ceiling height in the second case was estimated based on the pressure gradient in the first case. The FPG was found mainly generated by the contraction effect induced by the boundary layers over the tunnel walls because it can be largely removed in the third case. In addition, we investigated the mean and standard deviation of the surrounding wind field and surface pressure coefficient of an isolated square‐section building model. Because the FPG in the macroscale ABL is negligible compared with the adverse pressure gradient in the microscale near‐wall boundary layers of the building, the influence of the FPG is negligible in a wind tunnel experiment conducted in a relatively large wind tunnel with a safe blockage ratio.