Previous research has identified ethnic differences in adult obesity but has not fully explored the pathways that explain the ethnic differences, which may relate to both individual and area-level characteristics. This thesis identifies ethnic differences in obesity for eight key ethnic groups in England, before and after accounting for a range of individual-level and area-level factors. In addition, the thesis explores ethno-religious differences in obesity. Four key pathways to obesity are explored: social and structural inequality, migration-related, culture and health behaviour. Area-level factors examined are area deprivation, co-ethnic density and ethnic area type. As a basis for this comparative analysis, the thesis also examines approaches to the measurement of obesity and is the first study to use weight and waist as obesity outcomes in a study of ethnic differences in obesity (with height and hip respectively as explanatory variables). Data come from four years of the Health Survey for England (1998, 1999, 2003 and 2004) and linked area-level data from the 2001 Census. Multi-level modelling methods are used to account for individual-level and area-level factors. For waist, before adjusting for explanatory factors, Indian, Pakistani, Bangladeshi and Irish men and all seven of the women's ethnic minority groups have larger waists than the White group. After adjusting for individual and area-level factors, only the Indian and Pakistani men and Bangladeshi, Pakistani and Chinese women have statistically significantly larger waists than the White group. For weight, before adjusting for explanatory factors, Black Caribbean, Black African and Pakistani women are heavier than White women. However, in the adjusted models, the Pakistani women are no longer heavier than the White women. Both the unadjusted and adjusted models reveal that Black African Muslim women are lighter in weight than Black African non-Muslim women. Indian Sikh men and women have the largest waists of the Indian ethno-religious groups. The study finds that the pathway to ethnic and ethno-religious differences in obesity is multi-dimensional; a combined effect of social and structural inequality, migration, culture, and health behaviour for most ethnic groups. Waist circumference - individual level pathways: After adjusting for explanatory factors, the largest decrease to the waist circumference of men and women, relative to White women, is due to entering the migration-related variables into the models. This is consistent for men and women in all ethnic minority and all ethno-religious groups. In addition, entering the socio-economic position variables into the models substantially decreases the waist of Pakistani, Bangladeshi and Black African women, relative to White women - this is due to the low socio-economic position of these ethnic groups. Entering the socio-economic position variables into the models also decreases the waist of the Black African and Indian Muslim women to a greater extent than the non-Muslim Black African and Indian women. When the health behaviour variables are entered into the models, there is a large decrease in waist for Pakistani and Bangladeshi men relative to White men - this is due the low levels of physical activity among these ethnic groups compared to the White group. Entering health status into the models also results in a decrease to Bangladeshi men and women, Pakistani women, Indian women and Black Caribbean women's waist, relative to White women - this is due to the poorer health status of these groups. Weight - individual level pathways: After adjusting for explanatory factors, the largest decrease in women's weight, for all ethnic and ethno-religious groups relative to White women, is due to the migration-related variables and the health behaviour variables. For most groups the migration-related variables have the largest effect but for Pakistani and Bangladeshi women the health behaviour variables have the largest effect - this is due to the low levels of physical activity among Pakistani and Bangladeshi women. Entering the socio-economic position variables into the models also decreases the weight of the Black African and Indian Muslim women but increases (or makes little difference to) the weight of non-Muslim Black African and Indian women. In contrast, men's weight increases for all ethnic groups, relative to White men, after adjusting for all the explanatory factors - for all groups this effect is mainly due to the socio-economic position variables and, for the Indian and Pakistani men, the migration-related variables. Area-level pathways: Area-level pathways contribute to ethnic/ethno-religious differences in obesity but to a lesser extent than individual-level pathways. Predicted values from interaction models suggest that the effect of area deprivation on weight/waist is different for different ethnic groups. For example, White men and women's weight and waist increases, on average, as area deprivation increases but Indian men's waist and weight and Indian women's waist decreases as area deprivation increases. Interaction models also suggest that co-ethnic density may work differently for different ethnic groups - for example an increase in co-ethnic density results in a decrease to the weight or waist of the Indian, Chinese and Pakistani groups but an increase to the waist of Black Caribbean women. The ethnic area type research needs further study but the results suggest that Indian and Pakistani area types may have a protective effect upon both men's waist and weight.