Carbonated beverages have been reported to increase fracture risk in children but the mechanism is unclear. The aim of this population-based case-control study was to investigate the association between soft drink and milk consumption, physical activity, bone mass, and upper limb fractures in children aged 9-16 years. A total of 206 fracture cases and 206 randomly selected individually matched controls were studied. There were 47 hand fractures; 128 wrist and forearm fractures, and 31 upper arm fractures. An interviewer-administered questionnaire was utilized to retrospectively assess last-year physical activity (including television, computer, and video watching) and to recall the average weekly consumption of milk, colas, and total carbonated drinks. Bone mass at the spine, hip, and total body was assessed by dual-energy X-ray absorptiometry (DXA) and metacarpal morphometry. For total fractures, none of the above drink types was significantly different between cases and controls. For wrist and forearm fractures, there was a positive association between cola drink consumption and fracture risk (OR 1.39/unit, 95% CI: 1.01, 1.91). Cola consumption was significantly correlated with television, computer, and video watching (r = 0.20, P = 0.001) but not bone mineral density or milk drinks. After adjustment for television, computer, and video watching and bone mineral density, the association between cola drinks and fracture risk became nonsignificant (OR 1.31/unit, 95% CI: 0.94, 1.83). No association with other fracture sites was observed. In conclusion, cola, but not total carbonated beverage consumption, is associated with increased wrist and forearm fracture risk in children. However, this association is not independent of other factors and appears to be mediated by television watching and bone mineral density but not by decreased milk intake.