BACKGROUND: Given the limited research on behavioral determinants of using virtual exercise machines to engage in exercise during the COVID-19 pandemic, this study applied an integrated behavior change model to identify behavioral determinants using these machines. METHOD: Adult owners of livestreaming virtual exercise equipment (N = 123) completed measures of social cognition, planning, motivation, virtual exercise machine features, and sociostructural variables at an initial occasion (T1) and 4 weeks later (T2). Hypothesized relations among model constructs were tested using a cross-lagged structural equation model with past behavior and sociostructural variables as covariates. RESULTS: Autonomous motivation predicted intentions and habit indirectly via attitudes and perceived behavioral control at T1; virtual exercise machine features predicted intention and habit indirectly via social cognition constructs at T1; and intention and habit at T1 predicted exercise behavior at T2. CONCLUSIONS: This study supports social cognition constructs and virtual features as predictors of exercise using virtual exercise machines.