The worldwide prevalence of type 2 diabetes mellitus is increasing at a rapid rate, predominantly because of changes in environmental factors interacting with individual genetic susceptibility to the disease. Data from 20 longitudinal cohort studies present a consistent picture indicating that regular physical activity substantially reduces risk of type 2 diabetes. Adjustment for differences in body mass index between active and inactive groups attenuates the magnitude of risk reduction, but even after adjustment, a high level of physical activity is associated with a 20-30% reduction in diabetes risk. The data indicate that protection from diabetes can be conferred by a range of activities of moderate or vigorous intensity, and that regular light-intensity activity may also be sufficient, although the data for this are less consistent. The risk reduction associated with increased physical activity appears to be greatest in those at increased baseline risk of the disease, such as the obese, those with a positive family history and those with impaired glucose regulation. Data from six large-scale diabetes prevention intervention trials in adults with impaired glucose tolerance or at high risk of cardiovascular disease indicate that increasing moderate physical activity by approximately 150 minutes per week reduces risk of progression to diabetes, with this effect being greater if accompanied by weight loss. However, this level of activity did not prevent all diabetes, with 2-13% of participants per annum who underwent lifestyle intervention still developing the disease. Thus, while 150 minutes per week of moderate activity confers benefits, higher levels of activity may be necessary to maximize diabetes risk reduction in those at high baseline risk of the disease. In contrast, those at low baseline risk of type 2 diabetes, e.g. people with a very low body mass index and no family history of diabetes, will remain at low risk of developing diabetes whether they are active or not. Thus, the amount of physical activity required to confer low risk of diabetes differs according to an individual's level of baseline risk. Consequently, a 'one size fits all' mass-population strategy may not provide the most appropriate approach when designing physical activity guidelines for the prevention of type 2 diabetes. Producing tailored guidelines with the specific aim of reducing risk of diabetes in high-risk populations may provide an alternative approach.