There are many areas in medicine in which the diagnosis poses significant difficulties and depends essentially on the clinician’s ability to take and interpret the patient’s history. The differential diagnosis of transient loss of consciousness (TLOC) is one such example, in particular the distinction between epilepsy and ‘psychogenic’ non-epileptic seizures (NES) is often difficult. A correct diagnosis is crucial because it determines the choice of treatment. Diagnosis is typically reliant on patients’ (and witnesses’) descriptions; however, conventional methods of history-taking focusing on the factual content of these descriptions are associated with relatively high rates of diagnostic errors. The use of linguistic methods (particularly conversation analysis) in research settings has demonstrated that these approaches can provide hints likely to be useful in the differentiation of epileptic and non-epileptic seizures. This paper explores to what extent (and under which conditions) the findings of these previous studies could be transposed from a research into a routine clinical setting.