The EU criteria for the confirmation of the presence of illegal compounds in biological matrices were recently revised. The old and the revised criteria were applied to relative ion intensities obtained for five anabolic steroids (methylboldenone, methyltestosterone, ethynylestradiol, beta-boldenone and beta-nortestosterone) in meat (cow, pig, turkey) and fish at concentrations ranging from 0.5 to 5.0 microg/kg. Confirmatory analysis was done by GC-MS; therefore four diagnostic ions had to be monitored and three ion ratios had to be calculated and tested against the criteria. Application of the old and revised criteria, with either standards or fortified samples as reference, showed mutually rather divergent results. Confirmation according to the revised EU criteria and using fortified samples as a reference gave the best results; in other words the highest percentage of diagnostic ion ratios within the tolerance intervals. A correlation was found between the percentage of these ion ratios and the signal/noise (S/N) ratio of the least intense ion of interest in the recorded MS spectrum. Although there were distinct differences in the results obtained for different analytes and sample types, it is safe to conclude that at S/N=3 the percentage of ratios within the tolerance intervals generally will be at or below 50%, while for S/N>/=10, the percentage increases to over 90%. In the present study, fully satisfactory results were obtained down to about 2 microg/kg, but not for lower analyte concentrations.