The present research examines the process of the development of knowledge structures concerning events. Abelson (1976) suggested that one type of knowledge structure, a script, would move, as similar events were encountered, through three levels of generalization: from episodic, to categorical, and then to hypothetical. Specific hypotheses concerning the content of event knowledge structures at these varying levels of generalization were developed. Subjects were given descriptions of events to read. The number of events (one, two, three or four) and the level of similarity of the events (similar or dissimilar) were manipulated. Subjects were then asked to write what happened” in the material they had read. Responses to this question were content-analyzed. The results fit predictions. Subjects exposed to one event tended to write episodic protocols, full of concrete details about that event. Subjects exposed to two or three similar events, wrote somewhat more abstract protocols, with many of the characteristics of a categorical script. When exposed to four similar events, subjects’ protocols were most abstract, containing some of the characteristics of a hypothetical script. Less evidence of generalization was present when subjects read two, three, or four dissimilar events. The implications of these results were discussed, in terms of how the contents of inductive theories change as those theories are based on increasing amounts of supporting evidence.