Accurately predicting local ranges of isotopic signatures in human populations is essential for answering questions about past migrations and mobility. While local ranges of δ18O can be estimated using modern baseline samples and precipitation models, there are many environmental and anthropogenic drivers that can cause these ranges to deviate from the ranges seen in human populations. This study performs a geostatistical meta-analysis on a large dataset (n = 1,370) of spatially contextualized archaeological δ18O samples from 30 publications in order to generate a predictive model of local human δ18O ranges in the Central Andes. Two models were generated, one using archaeological samples of both humans and fauna, and the other using only humans. The model using only human samples makes more accurate predictions, cautioning against the incorporation of faunal δ18O samples in studies of human provenance. The models are also compared against a model of δ18O values found in precipitation across the study area, and significant differences lead to the conclusion that precipitation models are insufficient for predicting local human δ18O ranges.