This paper explores the use of portable Raman instruments to characterise natural dye lakes in paint mixtures, as an alternative approach to other Raman techniques (e.g. SERS). Raman spectroscopy has indeed been used extensively to study natural dyes as pure substances or as artistic pigments (dye lakes). However, the examination of these compounds with Raman spectroscopy is particularly challenging because of a strong fluorescence, a relatively weak Raman signal and their occurrence at low concentrations in artefacts. Because of these challenges, the typical way of analysing these materials is through either surface‐enhanced Raman spectroscopy (SERS) or Fourier‐transform (FT) Raman spectroscopy, yet those approaches are not always desirable in art analysis, especially as they often require micro‐sampling. Therefore, this study explores the potential of using commercial mobile instruments, which would open the possibilities for direct in situ analysis. Two dispersive instruments, one using a fibre‐optic probe and a 1064‐nm excitation laser and the other using the subtracted‐shifted excitation (SSE) post‐processing algorithm, have been tested in their feasibility to characterise dye lakes. Raman spectra were acquired from a set of laboratory reproductions of paint mixtures prepared with a chosen set of common lakes, three red (brazilwood, cochineal and madder) and one yellow (weld), mixed with natural proteinaceous and polysaccharide binders. The feasibility has been evaluated, and it is shown that these lakes produce a detectable Raman signal, in spite of the strong interference of the painting support (parchment) and that the two instruments provide significantly different information.