Physical activity (PA) during childhood and adolescence is important for the accrual of maximal peak bone mass, which helps to reduce the risk of fracture in later life. However, the dose of PA (amount, type, frequency, intensity) required to benefit bone remains unclear. This is likely due, in part, to a lack of standardised, comparable methods for assessing PA and development of methods that better target the measurements of bone-specific characteristics of PA in free-living populations. Accelerometers are frequently used to assess PA in free-living situations, however, common accelerometry methods are unsuitable for measuring short bursts of high-impact activity most relevant to bone. The thesis therefore aimed to develop and use accelerometry methods that more precisely quantify bone-relevant PA in children and adolescents. A systematic review of the accelerometry methods used in studies assessing associations between habitual PA and bone in children and adolescents was conducted (chapter 3). This demonstrated that the methods used varied widely between studies and the use of long epochs and cardiovascular-based intensity cut-points suggest that updated, bone-specific methods are needed. The next study (chapter 4) demonstrated that it was possible to adapt existing accelerometry methods (using 1-second epochs and smaller intensity increments) to identify a bone-specific PA intensity in a population-based sample of 11-12-year-olds. However, whilst these methods overcome some of the methodological limitations identified and will help to further understanding of the influence of PA on bone, it is likely that methods that use the raw acceleration signal (without an epoch), which has the resolution to identify detailed aspects of PA, (e.g. magnitude and number of impacts/loading) are needed in the future. Whilst more work is required before metrics directly derived from raw acceleration can be used in free-living situations, the final study in the thesis (chapter 5) demonstrated that peak magnitudes of raw acceleration from both wrist- and hip-worn monitors reflects the loading incurred during PA in this population and is therefore a suitable proxy measure of loading that can be applied to free-living situations in the future. The overall message in this thesis is that in order to establish the precise dose-response relationship between PA and bone, a shift in the approach to measuring bone-relevant activity is needed.