In cattle, carcass quality is an important criterion in the grid of payment to producers. The most accurate determination of beef carcass quality involves dissection of the entire carcass. This, however, is very costly and cumbersome. Alternative methods were developed from cut carcass or indirect measurements to calculate carcass composition using prediction equations: (1) dissection of the 9th, 10th and 11th rib followed by analysis of its chemical composition; (2) specific gravity of the whole carcass; and (3) combination of easily obtained measures (EOM) such as kidney-pelvic-heart fat, USDA yield grade (YG), quality grade, subcutaneous fat thickness (SFT), ribeye area, marbling and hot carcass weight. These alternative methods have become reference methods in research but have not been directly compared. Current objectives were to evaluate if the carcass composition calculated from the different reference methods matched. For this purpose, equations were developed between single proxy traits (YG, SFT) and reference carcass composition value, for each of the reference method. A meta-analysis was applied on 30 studies from 25 publications that reported carcass quality data from both proxy traits and reference methods (53, 31 and 16% of data based on rib dissection, specific gravity and EOM respectively). Within-study variance-covariance models were developed (Minitab 16) according to Sauvant et al. Results showed that the amounts of carcass lipids calculated from EOM are significantly lower than those calculated from rib dissection or specific gravity (-25 kg and -19 kg respectively, representing up to 17% of total carcass lipids). Similarly, amounts of carcass proteins calculated from rib dissection were significantly higher than those obtained from specific gravity or EOM (+3 kg and +2 kg respectively, up to 6% of total carcass protein). In conclusion, the three reference methods considered differ in their degree of accuracy to determine the chemical composition of the carcass.