Background The optimal distribution between physical activity (PA) levels and sedentary behaviour (SB) for the greatest benefits for body composition among older adults with overweight/obesity and chronic health conditions remains unclear. We aimed to determine the prospective association between changes in PA and in SB with concurrent changes in body composition and to examine whether reallocating inactive time into different physical activity levels was associated with 12-month change to body composition in older adults. Methods Longitudinal assessment nested in the PREDIMED-Plus trial. A subsample (n = 1564) of men and women (age 55–75 years) with overweight/obesity and metabolic syndrome from both arms of the PREDIMED-Plus trial was included in the present analysis. Participants were followed up at 6 and 12 months. Physical activity and SB were assessed using validated questionnaires. Out of 1564 participants, 388 wore an accelerometer to objectively measure inactive time and PA over a 7-day period. At each time point, participants’ body composition was measured using dual-energy X-ray absorptiometry (DXA). Standard covariate-adjusted and isotemporal substitution modelling were applied to linear mixed-effects models. Results Increasing 30 min of total PA and moderate-to-vigorous physical activity (MVPA) was associated with significant reductions in body fat (β − 0.07% and − 0.08%) and visceral adipose tissue (VAT) (− 13.9 g, and − 15.6 g) at 12 months (all p values < 0.001). Reallocating 30 min of inactive time to MVPA was associated with reductions in body fat and VAT and with an increase in muscle mass and muscle-to-fat mass ratio (all p values < 0.001). Conclusions At 12 months, increasing total PA and MVPA and reducing total SB and TV-viewing SB were associated with improved body composition in participants with overweight or obesity, and metabolic syndrome. This was also observed when substituting 30 min of inactive time with total PA, LPA and MVPA, with the greatest benefits observed with MVPA. Trial registration International Standard Randomized Controlled Trial (ISRCTN), 89898870. Retrospectively registered on 24 July 2014