Objective. To compare invasive and non-invasive techniques for measuring the posterior-to-anterior translations of vertebrae during spinal manipulative therapy. Design. This study represents a small part of a larger experiment. Background. Despite the mechanical nature of spinal manipulative therapy, the mechanism by which it alleviates back pain is still unknown. An understanding of the deformation behaviour of the spine during spinal manipulative therapy would aid in the formulation of a hypothesis underlying its efficacy. Methods. A clinician delivered posterior-to-anterior manipulative thrusts to the right transverse process of either T10, T11 or T12 in two cadavers. Posterior-to-anterior translations of the bone pins and surface markers (embedded in and taped over T10, T11 and T12, respectively) were recorded by cine cameras. Posterior-to-anterior accelerations, recorded by the accelerometers, were used to calculate posterior-to-anterior translations. Translation measurements made by using the surface markers and the accelerometers were compared to those made by using the bone pins. Results. There were no significant differences between the posterior-to-anterior translations of vertebrae obtained from the surface markers, as compared to the bone pins. The accelerometers underestimated the absolute, and overestimated the relative, vertebral translations, respectively, compared to the bone pins. Conclusions. The translations measured by the surface markers were more similar to the translations derived from the bone pins than those calculated from the accelerometers. Three-dimensional surface marker arrays would therefore be more useful to determine all relative movements non-invasively.