The determination of La and Ce anomalies in natural waters, biological samples and biogenic sedimentary rocks can provide unique information on biogeochemical processes in Earth surface environments. Over the last decades, several approaches have been used for calculating La/La* and Ce/Ce*, based on the comparison between measured and theoretical abundances (La* and Ce*) extrapolated from neighboring rare earth element concentrations normalized to chondritic or shale reference values. These extrapolations can be achieved either linearly or semi-logarithmically (“geometrically”), both methods being used in the literature in the absence of any consensus. We show here, using a database of rocks exhibiting no La and Ce anomaly, that the linear extrapolation of La and Ce abundances can result in markedly different results depending on whether chondritic or shale values are used for normalization. The geometric extrapolation allows consistent calculation of La and Ce anomalies for the entire compositional range tested in this study, regardless of whether data are normalized to chondritic or shale reference values. The differences between linear and geometric extrapolations are illustrated by a few selected examples from the literature, including various carbonate rock and seawater samples, further demonstrating that linear extrapolation can result in erroneous estimates of La and Ce anomalies. We thus propose that La/La* and Ce/Ce* ratios in all geological and environmental samples should be determined using the geometric extrapolation only.