Résumé : Traditional neurologic tenets claim that the clinical picture of acquired childhood aphasia is nonfluent irrespective of lesion location. In the past 20 years, however, several case studies have shown that fluent aphasic patterns can be observed in children with acquired childhood aphasia. But the question remains open as to whether the pattern of their speech characteristics is similar to the one described in adult aphasics as studies addressing spontaneous speech fluency characteristics in larger series of children with acquired childhood aphasia are scarce. The objective of this study was to investigate whether an analysis of spontaneous speech fluency as has previously been performed in adult aphasics by other investigators would also yield two distinct groups of aphasic children and, if so, whether the distribution of the different speech characteristics in both groups would reflect the rank order found in adults, that is, whether nonfluent verbal output characteristics would predominate in one group and fluent features in the other. Audiotaped and videotaped recordings of 24 cooperative children with acute acquired childhood aphasia unselected for age, gender, etiology, and aphasia severity ratings were analyzed according to 10 different speech characteristics. A cluster analysis (two-means clustering) was performed to seek the existence of two distinct groups of aphasic children. Results were confirmed, and exact P values were computed with Mann-Whitney U-tests. A two-means clustering created two distinct classes. Mann-Whitney U-tests ranked the speech characteristics according to their discriminating power between clusters. Comparing this rank order with the one previously found in adults revealed a high correlation (Spearman's rank correlation: r = .915, P << .005), thus indicating that the clusters we found were highly similar to the adult clusters. Thus, the use of the speech variables proposed to evaluate adult aphasic spontaneous speech enabled us to demonstrate a fluent/nonfluent dichotomy in a childhood aphasic population as well. This study shows that the traditional views on the uniformity of the clinical picture of acquired childhood aphasia are obsolete. Our findings corroborate data issued from several case reports of fluent acquired childhood aphasia and from the few studies focusing on speech fluency in acquired childhood aphasia, which all point to the existence of an adultlike heterogeneity of childhood aphasic syndromes. Current clinical evidence no longer supports the hypotheses of equipotentiality and progressive lateralization but favors the notion that the anatomic substrate for language representation in the child is similar to that in adults, even in young subjects.