Focal mechanism parameters of some significant earthquakes from southern Turkey have been estimated using either the body waveforms or the first motions of P-waves. It is observed that fault plane solutions derived from first motions may not be well constrained and may often be in error because of poor signal-to-noise ratios at some stations, poorly distributed recording stations in azimuth, lack of recognition of the reversed polarities at some stations, and positions of stations in terms of the crustal velocity structure, focal depth and, hence, take-off angles in the source area. The use of the inversion method introduces a considerable improvement in the focal parameters estimated in previous studies that have been used to construct seismotectonic models in the study area. The focal depths inferred from the inversion of body waveforms for the earthquakes occurring in southwestern Turkey are compatible with those reported by the ISC bulletins and other previous studies while the focal depths of earthquakes from the Antalya Bay are found to be less than those given by different agencies. Waveform inversion of earthquakes that have occurred in and around the Antalya Bay implies a low velocity layer showing azimuthal dependence. Hence, depth of rupture and lateral extent of velocity structure are of importance in the investigation of earthquake faulting process in this area. The source time duration of shallow earthquakes is generally longer than that for the deep focus earthquakes from the Antalya Bay and the Rhodes area. The events with relatively short and simple time functions may be interpreted as low stress drop events relative to the crustal ones. (C) 1999 Elsevier Science Ltd. All rights reserved.