We propose a method to assess the accuracy of atmospheric turbulence measurements performed by sonic anemometers and test it by analysis of measurements from two commonly used sonic anemometers, a Metek USA-1 and a Campbell CSAT3, at two locations in Denmark. The method relies on the estimation of the ratio of the vertical to the along-wind velocity power spectrum within the inertial subrange and does not require the use of another measurement as reference. When we correct the USA-1 to account for three-dimensional flow-distortion effects, as recommended by Metek GmbH, the ratio is very close to 4/3 as expected from Kolmogorov's hypothesis, whereas non-corrected data show a ratio close to 1. For the CSAT3, non-corrected data show a ratio close to 1.1 for the two sites and for wind directions where the instrument is not directly affected by the mast. After applying a previously suggested flow-distortion correction, the ratio increases up to approximate to 1.2, implying that the effect of flow distortion in this instrument is still not properly accounted for.