Conversational interaction between adults and mentally handicapped children is explored by analysing samples of naturalistic adult-child interaction obtained from three preschool mentally handicapped children playing with their mothers at home and their teachers in their classrooms. The adult input is analysed in terms of its grammatical structure and conversational functions, and comparisons are made between maternal and teacher input. The children's contributions to the conversational interaction are explored by analysing the words and word-like forms contained within prosodically defined utterances, focussing on the most productive child. The results reveal that whereas there were similarities in the maternal and teacher input, some differences emerge; for example, all adults took the major responsibility for the dialogue but, as a group, the teachers used a slightly higher proportion of well-formed grammatical utterances than the mothers who tended to make more use of short phrases, recitation and well learnt routines. Differences also occurred in the adults' use of language. The mothers, as a group, used higher proportions of their utterances to gain and maintain the children's attention and to request action rather than verbal responses. Although all the children were at the single-word stage, one of the children, Toby, made a larger contribution within the child- adult dyad than either of the other two children. Two aspects of his production are examined. First an analysis is made of his prosodic system and the association between certain prosodic units and communicative intent. Overall this association is found to be weak although there is evidence that he was marking two communicative functions. Secondly, detailed analyses are presented of his words and word-like forms. The range of Toby's vocabulary is similar to that reported for both normally developing children and other mentally handicapped children. However, contrary to reports in the literature, Toby is not restricted to the 'here and now' but is able to use his single words to comment on absent objects and events. These results have implications for intervention aimed at either the mother's input or vocabulary growth in the young mentally handicapped child.