The aim of this study was to compare thoracic spine alignment with two- and three-dimensional calculations of shoulder alignment (defined as a line joining the acromion processes of the right and left scapula) when all measures were projected onto the transverse plane. A six-camera Vicon system was used to reconstruct three markers positioned on the plane of the thorax such that the orthogonal vector to the thoracic spine, projected onto the transverse plane, was used as a virtual shoulder alignment during cricket fast bowling. This same measurement system was used to calculate the three-dimensional line between the acromion processes projected onto the transverse plane. These acromion markers were also used to calculate the two- dimensional transverse plane alignment of the shoulders from images recorded by a video camera positioned above ball release. All cameras operated at 50 Hz. A significant association was recorded between thorax alignment and the three- (0.97) and two-dimensional (0.87) shoulder alignment estimations at back-foot impact. The strength of association remained at front-foot impact, when correlations of 0.89 (three- dimensional) and 0.84 (two-dimensional) were recorded. However, at ball release, non-significant associations of 0.58 (three- dimensional) and 0.41 (two- dimensional), representing shoulder alignment differences of approximately 10degrees, were recorded. The 95% limits of agreement comparisons for shoulder alignment at back-foot impact, front-foot impact and ball release produced mean random errors for the two comparisons of 9.5degrees, 11.7degrees and 22.5degrees respectively. Three- and two- dimensional transverse plane projections of shoulder alignment are reasonable estimates of thorax alignment at back-foot impact and front-foot impact but not at ball release.