Various methods have been used to teach sight words to young children. Typically, the word to be taught is paired with an extra-stimulus prompt such as a picture depicting the word, and the child is presented with the word and the picture and taught to read the word through trial and error. However, there is some evidence which suggests that the use of extra-stimulus prompts in sight word acquisition may be inimical to the child's learning. This proposition was tested in a series of three experiments. In Experiment 1, the efficacy of an overcorrection procedure in teaching sight words to moderately mentally retarded children was established. This procedure was then used as the basic technique of instruction in the following two experiments. In Experiment 2, the effects of word-picture, word alone and a control condition on sight word acquisition was tested in an alternating treatments design. A similar comparison was undertaken in Experiment 3, except that the word-picture condition was replaced by a word-object condition. The results showed that overcorrection is an effective instructional technique for sight word acquisition, and that extra-stimulus prompts may indeed hinder the child's learning. These results were explained in terms of an attentional model of learning.