This paper provides an empirical comparison between two versions of a new accounting ratio devised by Davis (1990a, 1990b) and Kay (1993) and the traditional capital employed (ROCE) ratio. The new measures, 'added value on inputs employed' (AVIE) and 'added value on net output' (AVNO), are both based on a version of residual income they label 'added value', but differ in choice of scaling variable. Empirical evidence is provided in this paper on AVlE and AVNO, including cross-sectional patterns, distributional properties, sensitivity to alternative specifications, and rank-ordering and serial-correlation performance compared with ROCE. AVIE and AVNO differ markedly from ROCE in terms of absolute levels and account for at most half of the latter's cross-sectional variation. On the other hand, the results indicate that both the rank orderings and rank predictions of the different measures are very similar. The distributional properties of the added value ratios are sensitive to choice of scaling variable, suggesting that considerable care has to be exercised in drawing conclusions from the absolute rankings of these measures. The new added value ratios provide signals about firm rankings that differ only to a limited degree from those rendered by the traditional ROCE ratio.