The Rare Earth Elements (REE) archived in carbonate rocks retain a wealth of information on paleo-seawater chemistry and local-regional redox conditions. However, interpretations are often ambiguous due to the potential for REE remobilization in marine environments. In this regard, many carbonate rocks that retain primary seawater isotopic signatures exhibit non-seawater-like REE patterns, implying either unusual REE behaviour in seawater or diagenetic overprinting of otherwise well-preserved rock samples. Here, we apply sequential leaching to constrain the possible effects of different REE-bearing phases on measured carbonate REE patterns in order to address this quandary. Our results show that the exchangeable phase contains negligible REE, but could be an important host phase for other elements such as Sr, Ba and K. The acidified hydroxylamine hydrochloride leach is shown here not only to dissolve Fe–Mn oxides but also phosphate minerals, which induces middle REE enrichment in corresponding leachates of dolostone samples in our study. The demonstrable Fe–Mn oxide phase in limestone samples is characterized by middle and/or light REE enrichment and positive Eu anomalies, which are attributed to hydrothermal processes and continuing exchange with marine fluids after initial precipitation. The non-seawater-like REE patterns observed in the carbonate phase of otherwise well-preserved limestones resemble those of the co-existing Fe–Mn oxide fraction, and are interpreted to reflect the dissolution of REE carrier phases during early diagenesis. This view is supported by fluid-rock interaction modelling, which shows how REE can be mobilized at relatively low fluid/rock ratios in shallow porewaters. Non-seawater-like REE patterns may therefore be caused by the incorporation of REE from shallow porewaters before final lithification due to an elevated flux of particulate REE carrier phases, e.g. Fe–Mn oxides and organic matter, to the seafloor. In spite of the sensitivity of carbonate REE patterns to early diagenetic exchange, the co-occurrence of non-seawater-like patterns with primary strontium and carbon isotope values suggests that REE should not be viewed as a general indicator for the preservation of other geochemical proxies. Importantly, Ce anomalies of the carbonate phase will also be affected by porewaters, masking primary seawater values. Our results favour a reevaluation of redox interpretations by taking into account REE patterns as a whole.