Male inmates incarcerated in a Federal Correctional Institution viewed one of two videotaped scenes which were presented as recordings of a classification team meeting held in another institution. The two videotapes were actually simulations identical in content except with respect to the presence or absence of a segment in which an inmate model was shown to receive social pressure from his unit manager to enroll in group psychotherapy. This social pressure constituted a threat to the model's freedom to refuse therapy enrollment. It was predicted that observers of the freedom-threat tape would experience observation inductive reactance which would result in their desiring a greater amount of choice about group therapy enrollment at the same time that they would derogate the value of group therapy for themselves. Subjects also completed semantic differential profiles of the unit manager, inmate model, and themselves, allowing a test of predictions that the freedomthreatening unit manager would be rated as negative, potent, and active, while the threatened model would be seen as positive, less potent, and inactive. An optional boring task was presented to subjects to determine whether the freedomthreat observers would show a heightened tendency to refuse the task in order to indirectly reassert their freedom. It was also predicted that subjects with a history of frequent rules infractions would respond more strongly to the modeled freedom threat than would subjects with fewer infractions. The predictions regarding increased choice salience and increased derogation of group therapy for freedom-threat observers were supported by the data analysis. As anticipated, the unit manager was rated as negative and potent inthe freedom threat scene, while the threatened model was perceived as less potent than when he was viewed in the no-threat scene. An unexpected finding showed the model also to be evaluated as negative in the freedom-threat condition. Activity ratings of the unit manager and model did not differ significantly between threat/no threat observation conditions. Compliance rates on the boring task were too minimal in both conditions to permit a test of the indirect restoration of freedom hypothesis; this manipulation was therefore regarded as a failure. The study found only one effect for the individual-differences variable of prior rules infractions: subjects with a high number of infractions showed a greater tendency to negatively evaluate the freedom-threatened model. The results in general provide support for the theory that psychological reactance can be aroused by merely observing a threat to the freedom of a similar other. The negative evaluations of the model provided by freedomthreat observers may however, indicate that perceived similarity between the personal characteristics of the model and observer is not necessary in order for the effect to occur. Instead, the model's role, as similar to that of the observer, may be more important. It was suggested that cognitive consistency processes may need to be invoked in explaining the freedom-threat observers' derogation of group therapy.