Ice flow in divide areas is strongly anisotropic. The evolution of ice fabric, from the onset of divide flow towards steady state with a fully developed fabric, has been shown to profoundly affect both the stratigraphy and surface topography of ice divides. Here, we investigate the effects of ice flow on the age-versus-depth relationship at ice divides by using a full Stokes thermomechanical model with a non-linear anisotropic constitutive relation between stress and strain rates. We compare our results with analytical approximations commonly employed in age-depth predictions, such as the Dansgaard and Lliboutry approximations. We show that these approximations systematically underestimate the age of ice at fully developed divides by as much as one order of magnitude. We also show that divides with fully developed fabric are ideal locations for ice-core extraction because ice under them can be up to one order of magnitude older than ice at the same depth at the flanks. In addition, these divides have a distinctive morphological structure that allows them to be clearly identified from satellite imagery or ground-penetrating radar data.