Stimulus overselectivity is said to occur when behavior is under the control of a restricted set of stimuli from a stimulus complex. Three studies investigated the effects of specific multiple-stimulus training histories upon the overselective responding of normal preschool children. In Experiment 1, eight children 3 to 5 years of age, were trained to discriminate forms presented on cards. Each form was labelled with a nonsense syllable, and each card (multiple stimulus) consisted of two forms. A time-delay training procedure was used. Four subjects were trained using Concurrent training in which two of the three S- response choices contained components of the S+. Four subjects were trained using Sequential training in which the choices did not contain S+ components, but S+s were trained in order such that one component of a previously- trained S+ was present in the next S+. Subjects trained using Concurrent training acquired the discriminations in fewer trials, and had fewer errors during training. However, they responded to single components at chance level, whereas subjects trained using Sequential training recognized components and were able to recombine them into novel combinations. Experiment 2 was designed to investigate the effects of interspersing component probes among review trials of previously learned S+s that contained those components. Responses of two subjects to components were probed with interspersed review trials. Subjects with a Sequential training history demonstrated higher levels of correct responding to recombinations of components, whereas subjects with a Concurrent history continued to respond at chance level. In Experiment 3, two subjects with a history of Concurrent training were trained using Sequential training. The subjects learned to respond to recombinations and components at criterion level. It was concluded that multiple-stimulus training, in which S+ components are presented sequentially, is an effective method for training subjects to respond to components as well as to the total multiple stimulus. Findings are discussed in relation to: (a) attention theory and implications for related areas, such as stimulus salience and functionality; (b) a reassessment of the definition of overselectivity; and (c) implications for research with other populations.