The majority of Transtheoretical Model (TTM) research has employed a cross-sectional research design. While useful for characterizing participants within the various stages of change, it fails to capture the dynamic nature of change within the model (i.e., stage of change movement or transitional shift patterns). The purpose of this study was to characterize naturally occurring physical activity behavior change observed over 6 months among college students. Through this research we hope to identify variables actually associated with "change," as this might be especially useful in conducting future intervention research. This was accomplished using all four constructs of the TTM including stage of change, self-efficacy, processes of change, and decisional balance, along with the predisposing, reinforcing, and enabling factors from the PRECEDE/PROCEED model. Initially, female and male participants (N = 101) were grouped into transitional shift patterns on the basis of their self-reported stage of change results, which were collected at two time periods 6 months apart. The transitional shift patterns were derived from Cardinal, Engels, and Smouter's (2001) classifications scheme and allowed participants to be classified into groups labeled: stable sedentary, stable active, activity adopter, activity relapser, and perpetual preparer. Once classified into these transitional shift patterns, scores from the other constructs of the TTM obtained at baseline (i.e., first wave) helped determine the attributes that determined success or relapse in becoming physically active 6 months later. Variables from the first wave of self-report questionnaires were age, Body Mass Index, exercise behavior (in METS), self-efficacy (α = .88), decisional balance score (α = .72), cognitive process of change (α = .75), behavioral process of change (α = .89), predisposing factors (α = .64), enabling factors (α = .77), and reinforcing factors (α = .72). Multinomial logistic regression was used to compare several first wave scores across transitional shift patterns. This type of analysis was used because of its ability to compare across all transitional shift patterns and because each factor is controlled for while looking for unique contributions. It was revealed that exercise METs and the cognitive process of change differed across transitional shift patterns while decisional balance produced significant results for the activity adopters only. Both the theoretical and practical implications of these results will be discussed, as will directions for future TTM research. Funded by the Korean National Sport University and Korean Research Foundation. Keyword(s): exercise/fitness, health promotion, physical activity