A catalog for earthquakes that occurred in western Turkey during the period 1964-2010 is compiled for achieving homogeneity for magnitudes. Data are obtained from the International Seismological Center (ISC), where earthquake magnitudes are reported in different scales and come from a variety of sources. For establishing a common magnitude expression, namely an equivalent moment magnitude M*(w), new relations correlating the different magnitude scales with each other are derived from converting as many as possible of the magnitudes reported in the ISC bulletins. After magnitude conversions, the completeness magnitude M-c is sought by modifying the goodness-of-fit method of Wiemer and Wyss (2000) to become more appropriate for datasets with smaller sample size and higher Mc thresholds. The study region is divided into four smaller regions on the basis of spatial data homogeneity, while different periods of similar seismic network performance are recognized and tested to seek spatiotemporal variation of Mc. The results derived in each case are compared with those yielded by the application of both the original goodness-of-fit and maximum curvature methods and are found to be quite similar, although there are still cases with a difference exceeding 0.3 magnitude units. The goodness-of-fit method is very sensitive in the selection of the desirable percentage of fitting a power law (90% or 95%), whereas the proposed modification makes it independent of this level selection, and performing better especially for datasets that include events before 1990, when higher completeness magnitudes are evident.