The increase in proportion of the non-biological (D-) isomer of aspartic acid (Asp) relative to the L-isomer has been widely used in archaeology and geochemistry as a tool for dating. The method has proved controversial, particularly when used for bones. The non-linear kinetics of Asp racemization have prompted a number of suggestions as to the underlying mechanism(s) and have led to the use of mathematical transformations which linearize the increase in D-Asp with respect to time. Using one example, a suggestion that the initial rapid phase of Asp racemization is due to a contribution from asparagine (Asn), we demonstrate how a simple model of the degradation and racemization of Asn can be used to predict the observed kinetics. A more complex model of peptide bound Asx (Asn+Asp) racemization, which occurs via the formation of a cyclic succinimide (Asu), can be used to correctly predict Asx racemization kinetics in proteins at high temperatures (95-140 °C). The model fails to predict racemization kinetics in dentine collagen at 37 °C. The reason for this is that Asu formation is highly conformation dependent and is predicted to occur extremely slowly in triple helical collagen. As conformation strongly influences the rate of Asu formation and hence Asx racemization, the use of extrapolation from high temperatures to estimate racemization kinetics of Asx in proteins below their denaturation temperature is called into question. In the case of archaeological bone, we argue that the D:L ratio of Asx reflects the proportion of non-helical to helical collagen, overlain by the effects of leaching of more soluble (and conformationally unconstrained) peptides. Thus, racemization kinetics in bone are potentially unpredictable, and the proposed use of Asx racemization to estimate the extent of DNA depurination in archaeological bones is challenged.