Studies of infant feeding and weaning patterns in past populations that rely on a cross‐sectional approach must make the assumption that no infant mortality bias exists. Previous investigations of infant weaning patterns at the Dakhleh Oasis, Egypt, relied on cross‐sectional isotope data. In this study, we re‐examine this weaning pattern, using a simulated longitudinal approach, which does not require any assumptions regarding potential infant mortality biases. This involves examining the dental isotopic signatures of individuals who survived the weaning process. Stable isotope signatures from juveniles and adults (102 individuals, 297 teeth) were examined to reconstruct the weaning history of those that survived the weaning process. Both deciduous and permanent teeth were sampled. Homogenized enamel and dentin samples were isolated from each tooth and analyzed for δ¹³C_ap and δ¹⁸O_ap from the enamel and δ¹⁵N_coll and δ¹³C_coll from dentin collagen. We investigate differences between in utero versus postbirth, preweaning versus postweaning, and juvenile versus adult stable isotope values as reflected in the dentition. A random permutation procedure was used to test for statistically significant differences in stable isotope values between tooth types. Statistically significant differences were observed in all stable isotopes between permanent and deciduous teeth, and between early and later forming permanent teeth in δ¹³C_ap and δ¹⁵N_coll isotopes. These results indicate dietary change between in utero and postbirth, and changes occurring during the weaning period. These results provide a more comprehensive picture of infant weaning practices at Kellis and provide further support that complete weaning occurred by 3 years of age. Am J Phys Anthropol, 2007. © 2007 Wiley‐Liss, Inc.