Abstract69 beef roasts from different breeds, ages, growth histories and joints, were judged for texture, hot or cold, by a panel of 12 tasters. The panel was selected for discrimination and consistency over a series of training sessions during which a descriptive score sheet was developed through repeated discussion. Eight characteristics were decided upon to describe the texture of roast beef; seven on scales (resistance, wetness, juiciness, cohesiveness, hardness, overall texture, chew count) and one qualitative (adjectival descriptions).The panel average scores for the seven quantified characteristics were subjected to factor analysis and the loss of information involved in reducing the matrices to two dimensions was 6.3% (hot tasting) and 4.6% (cold tasting). Factor scores calculated for both sets of data agreed in being highly correlated (0.98>r>0.90); Factor 1 with ‘resistance’, ‘cohesion’, ‘hardness’, ‘overall texture’ and ‘chew count’ and Factor 2 with ‘wetness’ and ‘juiciness’, but poorly with any characteristic in the other group (0.11>r>0.03). Thus, the texture of roast beef was appreciated by this panel in two characteristics describable as ‘toughness‐tenderness’ and ‘juiciness’, and the more elaborate scoring system originally devised did not add appreciably to the precision of measurement.Comparison of factor scores on samples tasted both hot and cold showed that the relative toughness‐tenderness was maintained between samples under bot