The implementation of a new friction tension-only “GripNGrab” device attached to a rocking steel frame is described. The device, when subject to significant tension dissipates energy via sliding in the frictional component. When the device is loaded in the compression direction, almost no compressive force is carried, but displacement occurs in the ratchetting component. This absence of any significant compressive force within the dissipative system means that the rocking frame will always recentre after uplift from earthquake shaking. A 9 m tall 4.75m wide 3-storey steel concentrically braced rocking frame is designed for low-damage seismic performance. Restoring forces are provided by (i) gravity, (ii) friction “GripNGrab” (GNG) tension-only dissipation devices at the base, and (iii) beam-slab effects. The initial fundamental period of the structure was 0.16s. The initial structure used a 10mm GNG ratchet pitch, and had a GNG strength to not slide under serviceability level shaking. Elastic, pushover, cyclic pushover, as well as time history analyses, with different shaking intensities are conducted using OpenSEES software. The scope of work is limited to a single building and a single ground motion. Parameters varied included the presence of beam-slab effects, and the GNG device stiffness, strength and tooth pitch. It is shown that the full behaviour of the frame could be understood considering cyclic pushover analysis. The peak uplift displacement was conservatively estimated from the peak roof displacement using rigid body mechanics and the tension-only device provided no resistance to full frame recentring. For the frames considered, cumulative uplift displacements, necessary to determine the inelastic displacement capacity of the tension only device, were up to 28 times the peak uplift displacement, not necessarily occurring at the maximum shaking intensity. Maximum frame base shear force demands were up to 1.43 times that from pushover analysis. When the beam-slab, connecting the rocking frame to the rest of the structure, increased the lateral force resistance, the base shear increased significantly, reduced peak roof displacements, and increased the effective number of peak uplift displacement cycles (NPUDc). For large shaking intensities, yielding of the beam-slab occurred resulting in permanent peak roof and uplift displacements. The GNG device strength, stiffness and tooth pitch variations for the cases studied did not significantly affect the response. Initial stiffness, and secant stiffness, based methods to predict the response of rocking frames were non-conservative for these short-period structures with small energy dissipation, and a simple improvement to match the behaviour was developed for the case studied based on the R-T-m relationship for a range of shaking intensity.