Seven different nonlinear height (h)-diameter (d) models were developed and compared for brutian pine (Pinus brutia Ten.), black pine (Pinus nigra Arnold), and Taurus cedar (Cedrus libani A. Rich.) in southern Turkey. Residual analysis was conducted to identify the error structure. A weighting factor of w(i) = 1/d was found to be appropriate for achieving the equal error variance assumption. The performance of the models was compared and evaluated based on 6 statistical criteria and residual analysis. Results suggested that the Gompertz model was superior to the other models in terms of its predictive ability. These tree species were located throughout the Mediterranean region, covering a wide range of topographic and climatic conditions. It is well known that height-diameter relationships depend heavily on local environmental conditions. Differences in the height-diameter relationship among and between ecoregions were tested using the nonlinear extra sum of squares method. Test results suggested that height-diameter models significantly differed between ecoregions, indicating that ecoregion-based height-diameter models are needed for prediction purposes. The ecoregion-based height-diameter models developed in this study may provide more accurate information for developing forest growth and yield models.