The objective of this study was to assess the effect of animal types on the prediction (calibration) equations used in pig carcass classification. Paired comparison of these equations obtained from various animal types has been carried out by the authors for estimating the lean-meat percentage in pig carcasses (Y). The usual predictors have been used: carcass weight, two fat depth (F-1 and F-2) and one muscle depth (M) measurements on each carcass with the Danish Fat-O-Meater. In addition, the interaction term F-1 x F-2 was also included in the calibration equations. As might be expected, significant and often considerable differences were found between prediction (calibration) equations for separate sub-groups of pigs representing different crossbreeds and/or different fattening systems. Also, significant differences were obtained between three separate prediction equations and the joint, overall equation. However, in this case, the overall equation seems to give a still tolerable bias in comparing with the separate sub-group equations. Prediction equations, set up separately for castrates and females, were not significantly different within animal types. Several factors, inducing the instability of these prediction equations are also discussed in this paper. (C) 2003 Elsevier Ltd. All rights reserved.