The purpose of this study was to investigate the link between exercise behavior and social physique anxiety (SPA) among adolescents, and also to understand the SPA-exercise behavior relationship for controlling the demographic, psychological and behavioral variables. The participants were recruited from 18 high schools in Taiwan. Eight classes in each school were selected by convenience. A total number of 4,230 adolescents aged from 12 to 19 (2,223 boys and 2,007 girls) were participated, and completed the measurement of demographic variables (age, gender, social economic status [SES], weight and height), exercise behavior, weight loss, smoking, drinking behaviors, and psychological variables (SPA scale, self-esteem scale, contour drawing rating scale, and teasing experience). A three-step forced entry multiple regression analysis was performed to determine how much of the variance in SPA was explained by exercise behavior and other variables. Examination of the regression coefficient showed that exercise behavior was a significant and negative factor of SPA, accounting for 2.4% of the variance in SPA. Demographic variables were entered for the next step, adding a further 14.7% of the variance. Exercise behavior was significantly related to SPA. When behavioral and psychological variables were added to the model, the variance increased to 20.1%. The final model revealed that except SES, all other variables were significant factors, explaining 37.2% of the variance in SPA. The exercise behavior-SPA relationship remained statistically significant even after multivariate adjustment. This study further explored the exercise behavior-SPA relationship with a broad range of variables. This study concludes that exercise behavior accounts for a significant amount of unique variance in the presence of demographic, psychological and behavioral variables.