Dry-land training of the torso muscles is common in swimming programs; however, the role of the torso muscles in front crawl is unclear. The purpose of this thesis was to establish demands on the torso muscles in front crawl from analysis of kinematic data and torso muscle activity during sprint and middle-distance swimming. In the first two studies, 3D kinematic data were analysed from swimmers swimming at sprint and 400m pace. In Study 1, the range of hip roll decreased while shoulder roll range was similar as swimming speed increased. These differences produced greater range and velocity of torso twist as swimming speed increased, indicating higher demands on the torso muscles at sprint than at 400m pace. In Study 2, Fourier analysis was used to decompose angular momentum signals to determine the impact of the flutter kick on longitudinal body rotation. The third harmonic frequency, representing effects from the flutter kick, was greater in lower limb than in upper limb angular momentum at both paces, indicating a reduction in the rotation transferred from the lower to upper limbs. This reduction was more pronounced at sprint than at 400m pace. In Study 3, 3D kinematic data and surface EMG data from internal oblique, external oblique, rectus abdominis, and lumbar and thoracic erector spinae were collected from swimmers swimming at sprint and 400m pace. Torso twist angle did not relate to EMG data and no relationships between muscle activity and torso twist acceleration could be detected. Findings from this study indicate that the torso muscles may play a greater role in stability and posture than they do in producing torso twist in front crawl. Guidelines were developed to improve dry-land training specificity for swimmers: (1) increase torso muscle demands as swimming speed increases, (2) use the torso muscles to provide stability during lower limbs movements, and (3) challenge the torso muscles to maintain torso posture while moving the upper and lower limbs.