Simple negation in natural languages represents a complex interrelationship of syntax, prosody, semantics and pragmatics, and may be realised in various ways: lexically, morphologically and prosodically. In almost all spoken languages, the first two of these are the primary realisations of syntactic negation. In contrast, in many signed languages negation can occur without lexical or morphological marking. Thus, in British Sign Language (BSL), negation is obligatorily expressed using face-head actions alone (facial negation) with the option of articulating a manual form alongside the required face-head actions (lexical negation). What are the processes underlying facial negation? Here, we explore this question neuropsychologically. If facial negation reflects lexico-syntactic processing in BSL, it may be relatively spared in people with unilateral right hemisphere (RH) lesions, as has been suggested for other 'grammatical facial actions' [Language and Speech 42 (1999) 307; Emmorey, K. (2002). Language, cognition and the brain: Insights from sign language research. Mahwah, NJ: Erlbaum (Lawrence)]. Three BSL users with RH lesions were specifically impaired in perceiving facial compared with manual (lexical and morphological) negation. This dissociation was absent in three users of BSL with left hemisphere lesions and different degrees of language disorder, who also showed relative sparing of negation comprehension. We conclude that, in contrast to some analyses [Applied Psycholinguistics 18 (1997) 411; Emmorey, K. (2002). Language, cognition and the brain: Insights from sign language research. Mahwah, NJ: Erlbaum (Lawrence); Archives of Neurology 36 (1979) 837], non-manual negation in sign may not be a direct surface realisation of syntax [Language and Speech 42 (1999) 143; Language and Speech 42 (1999) 127]. Difficulties with facial negation in the RH-lesion group were associated with specific impairments in processing facial images, including facial expressions. However, they did not reflect generalised 'face-blindness', since the reading of (English) speech patterns from faces was spared in this group. We propose that some aspects of the linguistic analysis of sign language are achieved by prosodic analysis systems (analysis of face and head gestures), which are lateralised to the minor hemisphere.