Martinez-Jimenez, L (Martinez-Jimenez, Luis) Univ Talca, Escuela Fonoaudiol, Talca, ChileIntroduction. Language assessment in persons with brain injury makes it possible to know whether they require language rehabilitation or not. Given the importance of a precise evaluation, assessment instruments must be valid and reliable, so as to avoid mistaken and subjective diagnoses. Aim. To validate 'lexical screening for aphasias' in a sample of 58 Chilean individuals. Subjects and methods. A screening-type language test, lasting 20 minutes and based on the lexical processing model devised by Patterson and Shewell (1987), was constructed. The sample was made up of two groups containing 29 aphasic subjects and 29 control subjects from different health centres in the regions of Biobio and Maule, Chile. Their ages ranged between 24 and 79 years and had between 0 and 17 years' schooling. Tests were carried out to determine discriminating validity, concurrent validity with the aphasia disorder assessment battery, reliability, sensitivity and specificity. Results. The statistical analysis showed a high discriminating validity (p < 0.001), an acceptable mean concurrent validity with aphasia disorder assessment battery (r(s) = 0.65), high mean reliability (alpha = 0.87), moderate mean sensitivity (69%) and high mean specificity (86%). Conclusion. 'Lexical screening for aphasias' is valid and reliable for assessing language in persons with aphasias; it is sensitive for detecting aphasic subjects and is specific for precluding language disorders in persons with normal language abilities.