Background The authenticity of foodstuffs and associated fraud has become an important area. It is estimated that global food fraud costs approximately $US49b annually. In relation to testing for this malpractice, analytical technologies exist to detect fraud but are usually expensive and lab based. However, recently there has been a move towards non-targeted methods as means for detecting food fraud but the question arises if these techniques will ever be accepted as routine. Scope and approach In this opinion paper, many aspects relating to the role of non-targeted spectroscopy based methods for food fraud detection are considered: (i) a review of the current non-targeted spectroscopic methods to include the general differences with targeted techniques; (ii) overview of in-house validation procedures including samples, data processing and chemometric techniques with a view to recommending a harmonized procedure; (iii) quality assessments including QC samples, ring trials and reference materials; (iv) use of “big data” including recording, validation, sharing and joint usage of databases. Key findings and conclusions In order to keep pace with those who perpetrate food fraud there is clearly a need for robust and reliable non-targeted methods that are available to many stakeholders. Key challenges faced by the research and routine testing communities include: a lack of guidelines and legislation governing both the development and validation of non-targeted methodologies, no common definition of terms, difficulty in obtaining authentic samples with full traceability for model building; the lack of a single chemometric modelling software that offers all the algorithms required by developers. [ABSTRACT FROM AUTHOR]