This paper presents a methodology conceived as a support system to identify unknown materials by means of the automatic recognition of their Raman spectra. Initially, the design and implementation of the system were framed in an artistic context where the Raman spectra analyzed belong to artistic pigments. The analysis of the pigmentation used in an artwork constitutes one of the most important contributions in its global study. This paper proposes a methodology to systematically identify Raman spectra, following the way analysts usually work in their laboratory but avoiding their assessment and subjectivity. It is a three-phase methodology that automates the spectral comparison, which is based on one of the most powerful paradigms inmachine learning: the case-based reasoning (CBR) systems. A CBR system is able to solve a problem by using specific knowledge of previous experiences (well-known spectral library of patterns) and finding the most similar past cases (patterns), reusing and adapting them to the new problem situation (unknown spectrum). The system results in a global signal processing methodology that includes different phases such as reducing the Raman spectral expression by means of the principal component analysis, the definition of similarity measures to objectively quantify the spectral similarity and providing a final value obtained by a fuzzy logic system that will help the analyst to take a decision. The major benefit of a Raman spectral identification system lies in offering a decision-support tool to those who are not experts or under difficult situations with respect to Raman spectroscopy