Plant diagnostic laboratories are facing increasing demands for accreditation of their testing activities from both private and public organizations. For diagnostics, validating tests for the intended use is one of the key elements of accreditation. Thus far, validation has mainly focused on tests with a fixed scope, i.e. a test for organism X in matrix Y by method Z, suitable for the analysis of targeted samples. In many laboratories, however, targeted samples only form a minority of the samples. The majority consists of symptomatic (non‐targeted) samples including a huge variety of pest, host and matrix combinations. Since it is impossible to validate all tests used for the analysis of these non‐targeted samples, ensuring the reliability of the final results needs a different approach. This paper describes such an approach for diagnostic testing in plant virology, where the extent of validation of a test is determined based on the risks of a false result within the context of the diagnostic process. Examples illustrate how this risk‐based system approach relies on a tight inter‐relationship of validation, expertise and quality assurance within the diagnostic laboratory.