Introduction Conflicting evidence exists for associations between birth weight and childhood physical activity (PA) levels. It is important to know what other, potentially modifiable, factors influence PA in children given its' association with adiposity. Our aim was to identify predictors of childhood PA levels in the Gateshead Millennium Study (GMS), a population based cohort of 1029 infants born in 1999–2000 in Gateshead, Northern England.

Methods Throughout infancy and early childhood, detailed information was collected. Assessments at age 9 yrs included body composition, objective measures of habitual PA (using accelerometers during waking hours). Mean total volumes of PA (accelerometer count per minute, cpm) and moderate-vigorous intensity PA (MVPA), and the percentage of time spent in sedentary behaviour (% SB) were quantified and direct and mediating associations analysed within path models.

Results Significant differences were seen in all three outcomes between males and females (p<0.001). No direct significant associations were seen with birth weight. Increased paternal age was associated with significant increases in % SB and decreases in cpm and MVPA (p<0.033). Associations with BMI at 9 yrs were in the expected directions. Increased time spent in sports clubs was significantly associated with decreased % SB (p=0.02) and increased MVPA (p=0.01), but not cpm (p=0.13).

Conclusions Although we found no evidence for an effect of birth weight on PA, path models suggest indirect effects mediated through BMI. Having an older father appeared to have a negative impact on the child's PA levels, while participation in sports clubs increases time spent in MVPA, but not cpm.