The current research builds upon a previously published study that demonstrated the combination of Raman spectroscopy coupled with multivariate analysis (MVA) for the prediction of thermal maturity in coal by evaluating the efficacy of this method for the prediction of thermal maturity in shale. MVA techniques eliminate analyst bias in peak-fitting methods by using the full Raman spectrum, and then extricating the important spectral regions for distinguishing samples and building accurate, robust models. Partial least squares (PLS) regression models were developed using Raman spectra and VRo values (0.58–4.59%) for 53 geographically diverse shale chip samples, and 43 shale powder samples. Separate PLS models were built using Raman spectra from shale chips or powders. The calibration sets were validated using approximately one-third of the samples to rigorously assess the predictive accuracy of the models. The root mean standard error of prediction was 0.24 for the shale chip model, and 0.28 for the shale powder model. The coefficients of determination (R²) for the cross-validated data sets were identical (0.90, chips; 0.90, powders), revealing a strong linearity despite the geographic and age diversity of the samples. This study demonstrates the validity of using PLS models for the prediction of shale VRo from Raman spectra. The MVA method described herein presents a Raman alternative to the VRo industry benchmark for assessing thermal maturity in shale that is not imperiled by the shortcomings and subjectivity of peak-fitting methods.