Traditional obesity-related public health messaging often includes physical activity (PA) recommendations. However, at the population level, the data are conflicting, especially when comparing different self-reported vs measured techniques across different settings and populations. We measured the association between moderate-to-vigorous intensity PA (MVPA) and prospective weight change across five African-origin populations and the extent to which MVPA attenuated weight change over time. At baseline, 2,500 adults (median age: 37y) were recruited into the Modelling the Epidemiologic Transition Study (METS), from Ghana, South Africa, Jamaica, Seychelles, and US. 2000 participants were followed up 8 years later, with 851 participants having complete 7-day accelerometry to measure MVPA at both time points. Generalised estimating equations were used to explore the longitudinal association between weight and MVPA adjusted for several confounders. The obesity prevalence at baseline was 27.5% which increased to 38.0% at follow-up. Baseline MVPA varied from 7 (IQR: 4, 16) min/day in US women to 52 (IQR: 36, 78) min/day in South African men, and similarly at follow-up ranged from 8 min/day to 41 min/day among the same participant groups. While overall, engaging in higher MVPA levels was associated with a lower body weight, such that every additional 30 min of MVPA equalled a 600g lower body weight (p = 0.04), the interaction between time and MVPA was not statistically significant (p = 0.18). Therefore, regardless of the amount of MVPA at any time point, body weight increased over time. Despite the association between MVPA and weight, our results suggest that objectively measured longitudinal MVPA was not associated with the change in 8-year weight in African-origin adults. Our research confirms that while PA is a critical determinant of cardiovascular health, it alone may not be enough to stem the rising obesity burden.