Computed tomography (CT) gives accurate in vivo estimates of sheep carcass composition based on information provided by cross-sectional CT reference scans (Young et al., 2001). Spiral CT is a novel imaging technology in which contiguous cross-sectional scans of a known thickness are collected. This data allows the reconstruction of images in three dimensions, giving the possibility of a comprehensive assessment of characteristics that are defined in terms of shape, such as conformation and muscularity. The development of automatic procedures for image analysis is of high priority due to the large amount of information contained in the spiral scans. The first step in image analysis is isolating the carcass components of each cross-sectional image, called segmentation. An algorithm to automatically segment the cross-sectional images in the spiral scan was developed, with the objective of investigating in vivo assessments of conformation and muscularity in sheep. The aim of this study was to compare the muscle areas obtained by manual and automatic segmentations.