OBJECTIVES: To compare between accelerometry (MVPA-A) and self-reported activity (MVPA-SR) in activity-maintenance (Keep Active Minnesota; KAM) and weight loss-maintenance (Keep It Off; KIO) trials. METHODS: Linear regression estimated moderation of study, treatment, or time on MVPA-A and MVPA-SR associations. RESULTS: MVPA-A was similar between studies (KAM 119 minutes, KIO 112 minutes, p = .555), whereas MVPA-SR differed significantly (KAM 350 minutes, KIO 87 minutes, p < .0001). Only study moderated correlation between MVPA-A and MVPA-SR. MVPA-SR better predicted MVPA-A in KIO relative to KAM (p = .023). CONCLUSIONS: Results suggest that self-presentation bias may influence validity of self-report in intervention studies with activity as a primary outcome. Researchers should select self-report to assess activity dimensions that objective measures capture poorly.