Subjects were 138 inmates from the pre-release unit of a Southwestern prison system, randomly divided into three groups of 46 each. Each group viewed a video-taped model delivering a speech. The independent variable had three levels: (1) lecturer attired in a shirt and tie; (2) lecturer attired in a correctional officer's uniform; and (3) model attired in a T-shirt. After each group viewed the tapes, subjects were administered a knowledge test and an evaluation questionnaire. Results revealed significant effects on the knowledge test and on two questionnaire items. Inmates learned the most from a correctional officer model who was dressed in jeans and a T-shirt; they learned second best from a model attired in slacks, shirt and tie; they learned least from a model dressed in a correctional officer's uniform. Results support the following: (1) modeling behavior is an integral part of the characteristics of the model; and (2) modeling behavior is a result of the perceived degree of similarity between the observer and the model. In terms of observational learning theory, correctional agencies should discontinue the use of uniforms. (Author)