SUMMARYUnderstanding the crustal structure of the Anatolian Plate has important implications for its formation and evolution, including the extent to which its high elevation is maintained isostatically. However, the numerous teleseismic receiver function studies from which Anatolian Moho depths have been obtained return results that differ by ≤21 km at some seismograph stations. To address this issue, we determine Moho depth and bulk crustal VP/VS ratio (κ) at 582 broad-band seismograph stations, including ∼100 for which H–κ results have not been reported previously. We use a modified H–κ stacking method in which a final solution is selected from a suite of up to 1000 repeat H–κ measurements, each calculated using randomly selected receiver functions and H–κ input parameters. Ten quality control criteria that variously assess the final numerical result, the receiver function data set, and the extent to which the results are clustered tightly, are used to determine station quality. By refining Moho depth constraints, including identifying 182 stations, analysed previously, where H–κ stacking yields unreliable results (particularly in Eastern Anatolia and the rapidly uplifting Taurides), our new crustal model (ANATOLIA-HK21) provides fresh insight into Anatolian crustal structure and topography. Changes in Moho depth within the Anatolian Plate occur on a shorter length-scale than has sometimes previously been assumed. For example, crustal thickness decreases abruptly from >40 km in the northern Kirsehir block to <32 km beneath the Central Anatolian Volcanic Province and Tuz Golu basin. Moho depth increases from 30–35 km on the Arabian Plate to 35–40 km across the East Anatolian Fault into Anatolia, in support of structural geological observations that Arabia–Anatolia crustal shortening was accommodated primarily on the Anatolian, not Arabian, Plate. However, there are no consistent changes in Moho depth across the North Anatolian Fault, whose development along the Intra-Pontide and İzmir-Ankara-Erzincan suture zones was more likely the result of contrasts in mantle lithospheric, not crustal, structure. While the crust thins from ∼45 km below the uplifted Eastern Anatolian Plateau to ∼25 km below lower-lying western Anatolia, Moho depth is generally correlated poorly with elevation. Residual topography calculations confirm the requirement for a mantle contribution to Anatolian Plateau uplift, with localized asthenospheric upwellings in response to slab break-off and/or lithospheric dripping/delamination example candidate driving mechanisms.