Calculating crown formation times and enamel extension rates require the presence of unworn teeth. Unfortunately, most of the dental remains from archaeological sites are worn, so a reconstruction of the missing part is necessary. Here, we present a new application on fossil teeth of a recent methodology used to reconstruct minimally worn teeth based on tooth-specific regression equations (Modesto-Mata et al., 2017). Regression equations are based on the morphology of the cuspal area of unworn modern human molars, premolars, canines and incisors, both upper and lower. By defining three landmarks and fifty-one semilandmarks in specific microCT slices, we obtained XY coordinates which were used to calculate the regression models of every tooth type. To validate the application of these modern human regressions to fossil hominins, we selected six unworn teeth from Gran Dolina-TD6 and thirty-five teeth from Sima de los Huesos, both sites located in the Atapuerca hill (Spain). Then, using virtual imaging techniques, all these forty-one teeth were artificially worn until the dentine horns were exposed. They were subsequently reconstructed employing their tooth-specific regression equations and the reconstructed crown heights were measured. Lastly, we measured the real crown heights, which were then compared with the real ones. Our results show that the percentage error between the estimated and the real crown heights in Gran Dolina-TD6 and Sima de los Huesos teeth are less than 5%, ranging most of the errors between -2.5 and 2.5%. We consider values below 5% as acceptable, since it involves only a half decile of the cuspal region when we divide this distance into ten equal segments. Our results suggest that we can confidently employ these regression equations to reconstruct slightly worn hominin teeth. These regressions allow us to estimate percentage of lost enamel and to compensate number of perikymata in such area, making crown formation times and enamel extension rates usable for comparison purposes.