Three approaches to facilitating verbal and physical sharing and of the generalizability and durability of the behaviors that were trained were investigated. During a free play period, groups of preschool children were taught to share verbally, to share physically, or to share verbally and physically; another group was not trained. Immediately following free play, the children were observed in a different setting. Follow‐up was conducted 4 weeks after training ended. Physical sharing that was durable and generalizable resulted only when children were taught to share verbally. Increases in physical sharing produced by training children to share only physically were not durable and did not generalize. Training both verbal and physical sharing produced results with a magnitude slightly greater than teaching just verbal sharing. Despite a lack of special programming, some of the treatment effects generalized to another setting and were maintained during the Follow‐up. There was response generalization of the effects of training verbal sharing to physical sharing but not vice versa. Problems with the concept of response class, a methodological suggestion for studying response generalization, and possibilities concerning why generalization and maintenance occurred without specific programming are discussed.