In this paper, the statistical aspects of the methodology that led to the Meat Standards Australia (MSA) prediction model for beef palatability are explained and described. The model proposed here is descriptive: its intention is to describe the large amounts of data collected by MSA. The model is constrained to accord with accepted meat science principles. The combined dataset used in development of the prediction model reported is around 32 000 rows × 140 columns. Each row represents a sample tasted by 10 consumers; each column specifies a variable relating to the sample tested. The developed model represents the interface between experimental data, scientific evaluation and commercial application. The model is used commercially to predict consumer satisfaction, in the form of a score out of 100, which in turn determines a grade outcome. An important improvement of the MSA model relative to other beef grading systems is that it assigns an individual consumer-based grade result to specific muscle portions cooked by designated methods; it does not assign a single grade to a carcass.