Fifty beef animals varying in size, age and composition were used to determine the most accurate methods for determining beef composition. Separable and chemical composition of the 9-10-11th rib section and measurements of subcutaneous fat thickness were the most useful for explaining observed variability in carcass composition. For predicting percentage fat in fed steers, composition of the 9-10-11th rib section was the most accurate [coefficient of determination (R2 0·85)]. The 9-10-11th rib section was the most accurate and precise technique identified in this study to estimate beef carcass composition across and within all age classes. Specific gravity accounted for 68% of the variation in percentage of carcass fat across all age classes combined; however, specific gravity did not account for a significant portion of the variation within each age class. Deuterium oxide dilution was able to account for 77% of the variation in chemical fat percentage across all age classes combined; however, deuterium oxide dilution was not as effective in accounting for the observed variation in fat percentage within each age class. Ultrasound was found to be an accurate live measure of chemical fat percentage within all age classes combined.