We analysed the ground deformation produced by the M w = 6.1 2014 January 26 and M w = 6.0 2014 February 3 Cephalonia earthquakes, western Greece. Campaign GPS measurements and RADARSAT-2 synthetic aperture radar (SAR) interferometry provide constraints on the overall deformation produced by the sequence. TerraSAR-X and COSMO-SkyMed SAR interferometry provide constraints on the second earthquake separately. Two permanent GPS stations captured the two coseismic offsets and show no pre-or post-seismic transients. Most of the deformation is concentrated in the Paliki peninsula which is consistent with the location of the seismicity and the damages. Both GPS and SAR interferometry indicate areas with large deformation gradients probably due to shallow effects. Given the limitations on the data and on the knowledge of the structure and rheology of the crust, we used a simple elastic model to fit the ground displacements. Although such model cannot fit all the detail of the deformation, it is expected to provide a robust estimate of the overall geometry and slip of the fault. The good data coverage in azimuth and distance contributes to the robustness of the model. The entire sequence is modelled with a strike slip fault dipping 70 • east and cutting most of the brittle crust beneath Paliki, with an upper edge located at 2.5 km depth and a deeper edge at 8.5 km. This fault is oriented N14 • which corresponds to the azimuth of the Cephalonia Transform Fault (CTF). The fit to the data is significantly improved by adding a secondary shallow strike-slip fault with low dip angle (30 •) with a component of reverse faulting on that shallow fault. The modelling of the February 3 event indicates that the faulting is shallow in the north of Paliki, with a centroid depth of ∼3.2 km. The fit is improved when a single planar fault is replaced by a bent fault dipping ∼30 • in the uppermost 2 km and ∼70 • below. The fault of the January 26 earthquake, inferred from the difference between the two above models, is located south and beneath the February 3 fault, with a centroid depth of ∼6.4 km. We interpret the 2014 fault zone as an east segment of the CTF located ∼7 km east of the main axis of the CTF, which location is constrained by the elastic modelling of the interseismic GPS velocities. The aftershock sequence is mostly located between the January 26 fault and the axis of the CTF. According to our analysis, the Paliki peninsula is partly dragged north with the Apulian platform with ∼7 mm yr –1 of shear accommodated offshore to the west. During the last 30 yr three main sequences occurred along the CTF, in 1983, 2003 and 2014 breaking a large part of the fault, with a gap of 20–40 km left between Cephalonia and Lefkada.