Abstract Mammalian dental enamel is a key archive for the reconstruction of past environments. Sequentially mineralizing enamel provides continuous, several year-long records, which spatially-resolved sampling can ‘read’ at seasonal or higher time resolution. Yet it remains underexplored how much an initially incorporated compositional signal is overprinted by the two-stage enamel mineralization process, which affects the finally achievable time-resolution. We report results of a systematic investigation into histologically-defined compositional profiles from human enamel obtained by laser-ablation inductively-coupled-plasma mass spectrometry (LA-ICPMS). Sr/Ca, Zn/Ca, Ba/Ca and Pb/Ca are investigated as commonly-utilized proxies of (palaeo)diet, mineralization and/or pollution in modern and archaeological samples. Our modern human teeth are from unequivocally breast-fed and formula-fed individuals, respectively. By focusing on two profiles that are time-equivalent yet topographically different, namely the enamel-dentine junction (EDJ) vs. enamel prisms (P), we evaluate the compositional effect of enamel secretion vs. maturation on elemental proxies throughout enamel thickness. These two broadly orthogonal orientations are compared with the connecting Neonatal/Retzius (NNL/R) profiles that represent nominal isogrowth events during enamel secretion, which – if later maturation had no effect – should show invariant compositional signals across enamel. However, we find that all NNL/R profiles record systematically varying compositions across enamel, with Sr/Ca always decreasing by 30–80% and Zn/Ca increasing near-exponentially 20–35 times towards outer enamel. As such they can be used to benchmark the extent of enamel maturation. Since all prism profiles reveal signals similar to NNL/R for Sr, Zn and in part Ba, this unequivocally demonstrates that P-orientations are equally overprinted by maturation, in all cases with increasing intensity towards outer enamel. EDJ and P profiles do not match one-another despite representing coeval secretion time, and we infer that the highest degree of initial signal variability can be retrieved along the EDJ (approx. ≤ 100 μm) where maturation had the least effect. Using a simple two-component mixing model we show that during maturation only a moderate increase (1.3–1.9×) in discrimination against Sr is required to explain the Sr/Ca patterns. The Zn/Ca increase on the other hand is interpreted to reflect the Zn-binding motifs of the enzymes MMP-20 (matrix metalloproteinase-20) and KLK-4 (kallikrein-4) active during enamel secretion and maturation, which may preferentially imprint a Zn compositional signal in maturing enamel. In the case of Ba/Ca we find no systematic patterns analogous to that of Sr and Zn in modern samples. Moreover, cryptic diagenesis affecting Ba in otherwise well-preserved fossil teeth argues for caution when using Ba/Ca to infer nutritional signals without additional elemental/isotopic corroboration. Unlike all other investigated elements, Pb/Ca shows broadly similar EDJ-P profiles, and nearly invariant R-signals across enamel, which may reflect the affinity of Pb to both organic molecules and inorganic apatite during enamel mineralization. Overall our results reveal the element-specific behaviour of Sr, Zn, Ba and Pb during enamel mineralization.