​Arrays of bender elements are combined with simple, yet robust inversion algorithms to develop a device for S-wave tomography. A fixed frame configuration complemented with a new versatile bender element installation permits reducing measurement errors. System design involves optimal selection of transducer separation, a frame design that prevents wave transmission, and an adequate calibration procedure. Reliable tomographic images are obtained by combining data preprocessing and the regularized least squares solution. Given the small size of the data sets, inversion techniques based on a parametric representation of the medium are implemented as well. The tomographic system is tested at low confinement and within a true triaxial cell. Results show the potential of tomographic imaging in the characterization of geotechnical systems and in the monitoring of subsurface processes. In particular, shear wave velocity tomography permits monitoring changes in the velocity field, which is related to the average effective stress in freshly-remolded uncemented soils. A minimum anomaly size and velocity contrast are required for detection. Diffraction healing hinders the detection of low velocity anomalies.