Three major earthquakes (with Ms≥7) were clustered in space and time (1976--1984) near Gazli, Uzbekistan, without reported evidence for fault scarps. To constrain the kinematics of the seismic sequence, we invert both focal mechanism solutions of the mainshocks and major aftershocks and vertical surface displacements associated with the three major shocks. For seven M>4 earthquakes that occurred in the 1976--1984 period, we find that all the focal mechanisms agree with a single stress tensor. One of the mechanical implications of this mathematical solution is the choice of the seismic fault plane. The 1976 earthquake fracture planes are well constrained by the stress tensor to be striking E-W and ENE-WSW and dipping to the south and the NNW respectively. From stress tensor inversion alone, it is impossible to determine if the fracture plane of the 1984 event dips to the SE or to the NW. Coseismic vertical surface displacements allow us to infer possible fault plane orientations of the 1976 and 1984 events. The best fit to the surface displacement for the 1976 earthquakes is obtained from the geometries that are derived from the homogeneous stress tensor inversion. The inversion of vertical surface displacements for the 1984 event constrains the associated fault plane to correspond to a NE-WS striking and SE dipping plane. The inversions of complementary geophysical data argue for the complexity of the orientations of the faults activated by the 1976--1984 seismic sequence. The 1984 earthquake fault plane is well orientated to allow failure to be activated by an horizontal &sgr;1 under a Mohr-Coulomb failure criterion. Under the same failure criterion and stress conditions, unlike the 1984 fracture plane, the 1976 fault planes are misorientated and likely to correspond to preexisting discontinuities. Concerning the principle stress orientations, the Gazli region seems to be under control of the regional tectonic setting. From a regional point of view, the WNW-ESE strike of &sgr;1 and the NNE-SSW strike of &sgr;2 deduced from this study for the Kyzylkum Desert can be explained at once by the interaction of the regional plate boundary motions. ¿ American Geophysical Union 1996