Low level of physical activity (PA) has become an important public health problem even in low-income countries. The objectives of this study were to measure PA levels, determine the prevalence of low PA and identify socio-demographic factors associated with it in Bangladeshi adults. Data from 792 (urban, 395; rural, 397) Bangladeshi adults (25-64 years) were included in this population-based cross-sectional study conducted in 2011. Global Physical Activity Questionnaire version 2 (GPAQ-2) was used to measure PA. The metabolic equivalent task (MET) in minutes per week was calculated to determine total PA. Participants were categorized into low, moderate and high PA groups. Logistic regression was used to assess socio-demographic factors associated with low level of PA. Median MET-minute of total PA per week was almost double in the rural area (1720) than the urban area (960). The overall prevalence of low PA was 50.3% (95% CI: 46.8-53.8), urban 59.5% (54.7-64.3) and rural 41.9% (37.0-46.8). Women in general were more inactive (women 63.1% [58.3-67.9], men 39.3% [34.6-44.0]). The main contributions to total PA were from work (urban 40.0%, rural 77.0%) and active commute (57.0%, 21.0%). Leisure-time PA represented a very small proportion (<3.0%). Multiple logistic regressions found a significant association of urban residence (OR = 2.2; 95% CI: 1.5-3.2), women (2.1; 1.4-3.9), oldest age group 55-64 years (15.6; 7.5-32.2) compared to youngest age group 25-34 years, graduation or further education (8.6; 4.1-17.7), and higher socio-economic class (2.4; 1.4-4.2) compared to poor with insufficient PA. This study identifies low PA in a rural and urban population in Bangladesh and that further large-scale population studies are warranted.