In this paper we use Turkish household labor force data to address a number of conceptual issues pertaining to the wage curve, an empirically derived negative relationship between the real wage level and the local unemployment rate. First, we show that in developing economies where labor markets are prone to high degree of segmentation by skill level, local unemployment rates disaggregated by education provide more accurate measures of the degree of group-specific wage competition and hence yield more robust results of the wage curve analyses. Second, we estimate the wage curve using various definitions of the unemployment rate, including discouraged and marginally attached workers, and the long-term unemployment rate to explore the most relevant measure of local labor market tension in the wage setting process. We find that broader definitions of unemployment serve as a more effective reference point in measuring wage flexibility for women, whose attachment to the labor market is substantially weak in the Turkish context; while for men the official and long-term unemployment rates perform well. Finally, using quantile regression we show that wage responsiveness to unemployment cannot be assumed to be constant along the wage distribution. In the Turkish case, we find a higher unemployment elasticity of wages around the median segment of wage distribution. This effect is more pronounced for women. © 2013 Copyright Taylor and Francis Group, LLC.