Ground radiation measurements are necessary for all large solar projects and to evaluate the accuracy of solar radiation models and datasets. Ground-based data almost always contain low quality periods caused by instrumental problems, recording errors, or maintenance deficiencies. Therefore, quality control (QC) is needed to detect and possibly mark or exclude suspect or erroneous data before any subsequent analysis. The few existing automated QC methods are not perfect, so expert visual inspection of the data is still required. In this paper, we present a harmonized quality control procedure, which is a combination of several available methods, including some that provide for an expert visual inspection. Within the IEA PVPS 16 activity, these tests are applied to 161 worldwide stations equipped with various radiometer models, which are candidates for continuous benchmarking of irradiance datasets derived from satellite or meteorological models. Since the implementation of these methods and experts’ subsequent decisions could lead to different quality control results, the results obtained independently by nine evaluators are compared with reference to two test sites. Quality control results are similar and more rigorous than purely automated testing results, although some deviations exist due to differences in manual reporting.