Abstract: Even though wind-tunnel test data of the classical transonic airfoil RAE2822 have been widely exploited for validating numerical methods and CFD (computational fluid dynamics) solvers for decades, such validations still require special attention to uncertainties. These uncertainties mainly come from the computational grid, experimental data correction, geometry definition, correction and modeling, the definition of skin friction coefficient, and boundary layer velocity profile. Prior to validating CFD results, numerical methods need to be validated first. Especially, CFD results are supposed to be grid-independent. For this purpose, this paper develops a set of high-quality computational grids that avoid existing open-source grids' deficiencies, yielding expected grid convergence. Moreover, the fidelity of the adopted CFD solver in the present study is cross verified with other solvers. When comparing CFD and experimental results, it is necessary to correct Mach numbers and (or) angles of attack. Although effective, the camber correction is not universal but needs to consider flow conditions such as Mach number, angle of attack, and Reynolds number. A limited number of discrete points defines the original airfoil geometry. Therefore, it is inevitable to interpolate the geometry to generate grids. However, different interpolation methods show minor effects on the numerical simulation of the flow field around the airfoil leading edge, where the curvature is the largest. Most relevant CFD research work only compared pressure distributions with wind-tunnel test data; A few studies compared the frictional resistance coefficient, boundary layer and wake velocity profile, but in comparison, it is necessary to pay attention to the difference between the definition of relevant parameters of the original wind tunnel test and the common definition of CFD calculation.