Purpose: : To explore among demographic groups the differences in prevalence estimates of physical activity that may occur as a result of differences in survey design characteristics, including question wording, placement, and examples of activities.Methods: : We compared responses to similar physical activity instruments administered to large samples of adults in 1999 (n = 9,775), 2000 (n = 32,374), and 1999–2000 (n = 7,529). The questions assessed participation in non-occupational physical activity at vigorous and moderate intensities. Surveys used in-person or telephone interviews.Results: : The prevalence of recommended levels of physical activity (i.e., ≥3 days and ≥20 min vigorous activities or ≥5 days and ≥30 min moderate activities) varied 10% across 3 surveys. Although survey design characteristics varied, higher prevalence was associated with the use of examples to measure multiple domains of activity and question order.Conclusions: : Measuring multiple domains is important for assessing health-related physical activity. These results suggest that physical activity measurement varies with question and survey design characteristics.