ObjectivesTo explore the association of dietary behaviour factors on obesity among city, township and rural area adults.SettingA stratified cluster sampling technique was employed in the present cross-sectional study. On the basis of socioeconomic characteristics, two cities, two townships and two residential villages were randomly selected where the investigation was conducted.ParticipantsA total of 1770 city residents, 2071 town residents and 1736 rural area residents participated in this survey.Primary and secondary outcome measuresDietary data were collected through interviews with each household member. Anthropometric values were measured. Participants with a body mass index (BMI) of ≥28.0 kg/m2 were defined as obesity.ResultsThe prevalence of obesity was 10.1%, 7.3% and 6.5% among city, township and rural area adults, respectively. Correlation analysis showed that for adults living in cities, the daily intake of rice and its products, wheat flour and its products, light coloured vegetables, pickled vegetables, nut, pork and sauce was positively correlated with BMI (r=0.112, 0.084, 0.109, 0.129, 0.077, 0.078, 0.125, p<0.05), while the daily intake of tubers, dried beans, milk and dairy products was negatively correlated with BMI (r=−0.086, −0.078, −0.116, p<0.05). For township residents, the daily intake of vegetable oil, salt, chicken essence, monosodium glutamate and sauce was positively correlated with BMI (r=0.088, 0.091, 0.078, 0.087, 0.189, p<0.05). For rural area residents, the daily intake of pork, fish and shrimp, vegetable oil and salt was positively correlated with BMI (r=0.087, 0.122, 0.093, 0.112, p<0.05), while the daily intake of dark coloured vegetables was negatively correlated with BMI (r=−0.105, p<0.05).ConclusionsThe prevalence of obesity was higher among city residents than among township and rural area residents. The findings of this study indicate that demographic and dietary factors could be associated with obesity among adults. Healthy dietary behaviour should be promoted and the ongoing monitoring of population nutrition and health status remains crucially important.