Context The mechanisms underlying regional differences in growth parameters of common carp, Cyprinus carpio Linnaeus, 1758 (Actinopteri: Cyprinidae), are not yet fully understood, because comparisons among them have been made post hoc. Aims To simultaneously compare growth of common carp between sexes and between regions: Demirköprü Reservoir (DR) and Marmara Lake (ML), which have different structures (lowland type vs highland type) and are proximate to each other in the same river basin (Gediz River Basin). Methods Length-frequency distributions, length–weight relationships and three growth models (von Bertalanffy growth function (VBGF), Gompertz growth function (GGF) and Robertson growth function (RGF)) were used for both sexes and all individuals in reservoirs. Key results As none of the best-fitting models used among these growth models clearly suggested a ‘clear winner’ (Akaike’s weight (wi) > 0.9), the model-averaged multi-model inference (MMI) was used for the first time for common carp. Conclusions Asymptotic length obtained from MMI for all individuals in ML (lowland type) was statistically higher than that in DR (highland type) (P < 0.05). Implications Based on the maximum observed length and the mean L∞ obtained from MMI, the common carp population in the DR can be considered threatened by overfishing.