The purpose of this project was to (a) compare traditional and virtual modeling when learning a multiple degree-of-freedom skill (fly casting), and (b) determine if the presence or absence of an authority figure would affect learning via virtual modeling. Sixty-nine undergraduate students with no previous experience were randomly assigned to one of four groups. The participants were asked to fly cast a lure into a target area at a distance of 35 feet. Day 1 consisted of a 10 trial pre-test and during Day's 2-6 participants practiced 80 casts and tested on trials 81-90. They were scored on accuracy, aesthetics of the throw, and skill acquisition . The model was viewed either live or in a virtual environment after every 20 trials for the first two treatment days, and every 40 trials for Days 3-6. Five days after the final treatment and testing session the participants were involved in a retention test. Group 1 (n=17) viewed a virtual model embedded in a multimedia software (with an authority figure present) of the appropriate casting mechanics as demonstrated by a trained and expert model ; Group 2 (n=16) viewed a virtual model embedded in a multimedia software (with no authority figure present) of the appropriate casting mechanics as demonstrated by a trained and expert model; Group 3 (n=17) viewed a live trained and expert model; and Group 4 (n=19) acted as the control. ANOVAs confirmed no significant differences between groups on the pretest for Accuracy Scores F(3,68) = .235, p = .872, Aesthetic Quality Scores, F(3,68) = .992, p. = .402, and Skill Acquisition Scores, F(3,68) = 1.136, p = .341. ANOVAs with repeated measures were used to compare group performance differences across time from pretest to posttest. These analyses indicated significant differences in performance for Accuracy Scores F(3,68) = 2.873, p = .001, Aesthetic Quality Scores F(3,68) = 2.364, p = .013, and Skill Acquisition Scores F(3,68) = 4.741, p = .001. Scheffe post-hoc analyses showed significant differences between Group 4 and all other groups for all three dependent measures. Paired-sample t-tests indicated that learning had occurred for Groups 1, 2, and 3 on all three dependent measures. The results suggest that all three treatments resulted in significant learning for fly casting, and virtual modeling with or without an authority figure present is as effective as a live model.Keyword(s): . NA