The aim was to test a new instrument, the BACES system, an articulated arm connected to a computer, to improve noninvasive examination of the spine and to reduce X-ray examinations. To this aim, four observers performed three series of eight measurements on a dummy and two healthy volunteers. Data on the projection of the spine, any curve detectable on the frontal and sagittal plane, and any rotation at each thoracolumbar level were collected and analysed. To detect the role of the observer, of the procedure and of the object in measurement variability, a two-factor analysis of variance was performed. The results showed measurement errors for kyphosis and scoliosis generally below 3 degrees, whereas the lordosis showed a major variability, especially in volunteers, because of the attitude role. Rotations showed a limited variability too, generally below 2 degrees. In the clinical setting, > 2 SD may reasonably be considered a clinical change because it has not a great chance of being a measurement error. Thus, the BACES system allowed us to identify with fairly good precision kyphosis, scoliosis and trunk rotations. Even though at the moment no surface measurement method can replace X-rays, the instrument can integrate and limit the use of radiographic evaluations.