Australian sheep genetic evaluation is conducted routinely for millions of animals for many traits. In the current analysis implemented by the OVIS software, phenotypes are pre-adjusted for systematic fixed effects to make fair genetic comparison between animals. This study assessed whether correction factors used in OVIS remain valid, and to explore whether the pre-adjustment method is still suitable and is comparable with a linear model. Furthermore, importance of interactions between body weight and sex, or body weight and flock were estimated. Regression slopes were calculated from forward prediction, using eye muscle depth data on 234,810 White Suffolk and 249,136 Poll Dorset sheep and fat depth data on 246,149 White Suffolk and 268,002 Poll Dorset sheep. Updated pre-adjustment factors produced regression slopes of progeny performance on their sire’s estimated breeding values (EBVs) equal to 0.67 and 0.62 (averaged over breeds) for eye muscle depth and fat depth, respectively. Regression slopes were same for eye muscle depth and slightly better for fat depth than OVIS (0.66 and 0.64 respectively). A linear model produced significant improvements in regression slopes (0.60 and 0.50 respectively). Including interaction effects between fixed effects did not significantly influence the accuracy of prediction of progeny performance. A linear model will be implemented in future OVIS evaluation for ultrasound scan carcass traits.