Naturally occurring and deeply coloured iron-bearing materials were exploited very early on by human populations. The characterization of these materials has proven useful for addressing several archaeological issues, such as the study of technical behaviors, group mobility, and the reconstruction of cultural dynamics. However, this work poses some critical methodological questions. In this paper, we will review ochre studies by focusing on the analytical methods employed, the limits of non-invasive methods, as well as examples of some quality research addressing specific issues (raw material selection and provenience, heat treatment). We will then present a methodological approach that aims to identify the instrumental limits and the post-depositional alterations that significantly impact the results of the non-invasive analysis of cohesive ochre fragments from Diepkloof rock Shelter, South Africa. We used ochre materials recuperated in both archaeological and geological contexts, and we compared non-invasive surface analyses by XRD, scanning electron microscopy coupled with dispersive X-ray spectrometry (SEM-EDXS), and particle-induced X-ray emission (PIXE) with invasive analysis of powder pellets and sections from the same samples. We conclude that non-invasive SEM-EDXS and PIXE analyses provide non-representative results when the number of measurements is too low and that post-depositional alterations cause significant changes in the mineralogical and major element composition at the surface of archaeological pieces. Such biases, now identified, must be taken into account in future studies in order to propose a rigorous framework for developing archaeological inferences.