The use of a traditional syndrome-classification system for aphasics is examined critically from the different perspectives of medicine and psychometrics. Medicine views syndromes as dichotomous (present or not present) and necessarily indicative of an underlying pathognomic state or process, and psychometrics sees performances as varying along a continuum. The issues of the pragmatic value of syndromes, or their correctness versus their fruitfulness, and the procedures for identifying syndromes are discussed in the context of recent research. The use of distinctive recovery patterns as a means of classifying aphasics and determining the appropriate therapeutic approach for each is proposed and illustrated with a model. (MSE)