Understanding the progression of physical activity behaviour from childhood to adulthood requires a valid, reliable and practical method of assessing activity levels which is appropriate for use in large groups. The measurement of physical activity in large scale research projects requires a method which is low in cost, agreeable to the study volunteer and accurate. Self-report can be used to determine adult activity patterns, but children lack the cognitive ability to recall details about their activity patterns. Heart rate telemetry has been used to estimate daily activity in children as a sole criterion and to validate commercial accelerometers. However, heart rate is an indirect estimate of physical activity which makes assumptions based on the linear relationship between heart rate and oxygen uptake. It is sensitive to emotional stress and body position, and takes longer to reach resting levels after physical exertion compared with oxygen uptake. It also lags behind movement, particularly as children's physical activity is spasmodic or intermittent in nature. One alternative is the pedometer. Many early studies reported that the pedometer is inaccurate and unreliable in measuring distance or counting steps. While reasonably accurate at mid range speeds, the accuracy of the pedometer decreases in very slow walking or very fast walking or running. However, more recent studies have examined the efficacy of using pedometers to assess daily or weekly activity patterns as a whole, and these have produced more promising results. In this regard, the pedometer has a number of advantages. It is very cheap, objective and does not interfere with daily activities and is therefore appropriate for use in population studies. Commercial accelerometers with a time-sampling mechanism offer further potential and could be used to provide a picture of the pattern of children's activity. As it has been observed that prolonged activity periods are not typically associated with childhood behaviour patterns, the use of a threshold value for 'aerobic' training stimulus is not appropriate as a cut-off value for physical activity. Instead, there is evidence to suggest that the total activity data measured by pedometers over limited periods of time may be more appropriate to assess how active children are.