<p>The analysis of samples by using spectroscopic fingerprinting techniques is more and more common and widespread. Such approaches are very convenient, since they are usually fast, cheap and non-destructive. In many applications no sample pretreatment is required, the acquisition of the spectrum can be performed in about one minute and no solvents are required. As a consequence, the return on investment of the related technology is very high.<br />The "disadvantage" of these techniques is that, being the signal non-selective, simple mathematical approaches (e.g., Lambert-Beer law) cannot be applied. Instead, a multivariate treatment must be performed by using chemometrics tools.<br />In what concerns food analysis, they can be applied in several steps, from the evaluation of the quality and the conformity of raw material to the assessment of the quality of the final product, to the monitoring of the shelf life of the product itself. Another interesting field of application is the verification of food-authenticity claims, this being extremely important in the case of foods labeled as protected designation of origin (PDO), protected geographical indication (PGI) and traditional speciality guaranteed (TSG).<br />In the present paper, it is described how non-selective signals can be used for obtaining useful information about a food.</p>