Aims:A recent catalogue by Flesch & Hardcastle presents two major anomalies in the spatial distribution of QSO candidates: i) an apparent excess of such objects near bright galaxies, and ii) an excess of very bright QSO candidates compared to random background expectations in several regions of the sky. Because anyone of these anomalies would be relevant in a cosmological context, we carried out an extensive analysis of the probabilities quoted in that catalogue. Methods: We determine the nature and redshift of a subsample of 30 sources in that catalogue by analysing their optical spectra (another 11 candidates were identified from existing public databases). These have allowed us to statistically check the reliability of the probabilities QSO status quoted by Flesch & Hardcastle for their candidates. Results: Only 12 of the 41 candidates turned out QSOs (7 of which have been identified here for the first time). Conclusions: The probabilities of the QSOs' being the candidates given by Flesch & Hardcastle are overestimated for mB ≤ 17 and for objects projected near (≤1 arcmin) bright galaxies. This is the cause of the anomalies mentioned above.