We aimed to assess the agreement of a commercially available bioelectrical impedance analysis (BIA) device in measuring changes in fat, lean and bone mass over a 10-week lifestyle intervention, with dual energy X-ray absorptiometry (DXA) as reference. A sample of 136 volunteers (18-66 years) underwent a physical activity intervention to enhance lean mass and reduce fat mass. BIA (Tanita BC545) and DXA (Hologic Explorer) measures of whole-body composition were taken at baseline and at the end of the intervention. After an average of 74 ± 18 days intervention, DXA showed significant changes in 2 of 3 outcome variables: reduced fat mass of 0.802 ± 1.092 kg (P < 0.001), increased lean mass of 0.477 ± 0.966 kg (P < 0.001); minor non-significant increase of 0.007 ± 0.041 kg of bone mass (P = 0.052). The respective changes in BIA measures were a significant reduction of 0.486 ± 1.539 kg fat (P < 0.001), but non-significant increases of 0.084 ± 1.201 kg lean mass (P = 0.425), and 0.014 ± 0.091 kg bone (P = 0.074). Significant, but moderately weak, correlations were seen in absolute mass changes between DXA and BIA: 0.511 (fat), 0.362 (lean) and 0.172 (bone). Compared to DXA, BIA demonstrated mediocre agreement to changes in fat mass, but poor agreement to lean mass changes. BIA significantly underestimated the magnitude of changes in fat and lean mass compared to DXA.