The Transtheoretical Model (Prochaska & Marcus, 1994) has been widely used as a framework for understanding exercise behavior change. This study's purpose was to clarify equivocal research findings reported when examining stage movement over time rather than static stages. Longitudinal studies, while providing modest support for related model constructs over time (Pinto et al., 2001), have also suggested that some constructs may not operate as predicted by the model (Nigg, 2001; Plotnikoff, 2001). These mixed findings may be due to length of time examined, combined transitional stages, and examination of only early stage movement. Participants were 528 female (n = 285), and male (n = 243) undergraduates (M age = 19.9, SD = 2.7). Participants completed previously validated questionnaires assessing stage of change, exercise behavior, processes of change, pros and cons of exercise, and exercise self-efficacy, two times, separated by 9 weeks. Participants were classified into one of five transitional shift groups based on their responses to the stage of change questionnaire at baseline and follow-up: (a) stable sedentary, (b) stable active, (c) activity adopters, (d) activity relapsers, and (e) perpetual preparers. Results of a 5 (group) X 2 (time) RM ANOVA examining exercise behavior revealed a significant interaction (p < .001), and supported transitional stage classification, with activity adopters and stable actives increasing exercise over time and relapsers decreasing activity. Separate 5 (group) X 2 (time) RM ANOVAs examining model constructs revealed no significant interaction for cognitive processes of change, however activity adopters and stable actives reported significant (p < .01) increases in the use of behavioral processes over time, while only the activity relapsers and perpetual preparers reported decreases. Activity relapsers also reported significant (p < .05) decreases in the pros of exercise. No significant interactions were found for the cons of exercise behavior, or for exercise self-efficacy. Findings supported the greater importance of behavioral rather than experiential processes in stage movement. No support was found for model predictions that pros of exercise should increase during activity adoption, or for the role of cons of exercise. Unlike findings reported in cross-sectional studies, increases in self-efficacy did not accompany increases in exercise stage. Nigg (2001) reported similar findings, perhaps supporting the notion of reciprocal determinism (Bandura, 1986). Finally, current study findings provide strong support for examination of stage movement classifications rather than static stages, as these transitions provide greater insight into the mechanisms of exercise behavior change.Keyword(s): exercise/fitness, health promotion, physical activity