Studies find that longer-term immigrants have higher body mass index (BMI) than their more recently-arrived counterparts. Most interpretations of these health patterns by duration of U.S. residence rely on theories of immigrant integration; they posit that with increasing time in the United States, immigrants incorporate economically, socially, and culturally into aspects of U.S. society, and that these changes impact health. Few studies empirically examine whether these aspects of integration are indeed mediators of the association between duration of U.S. stay and BMI, and if their patterns differ across immigrant subgroups. This study examines data from the National Latino and Asian American Survey, using path analytic methods to simultaneously test six hypothesized mediators between duration and BMI: household income, English language ability, ethnic identity, family cohesion, acculturative stress and discrimination for both Latino and Asian immigrants, stratified by gender. We find little evidence for an association between duration and BMI for either Latino or Asian men. For women, duration and BMI have a significant and positive relationship, although the pathways differ between the two ethnic groups.For Latina women, household income and acculturative stress are significant indirect pathways, although they work in opposing directions. For Asian women, English proficiency and discrimination are significant indirect pathways. Our findings reveal complex pathways between duration and BMI that vary by ethnicity and gender and highlight limitations in the negative acculturation theory, which suggests that exposure to the United States should have a net negative impact on health. In contrast, our findings suggest not all groups show declining health with longer duration, as measured by BMI, and that integration processes do not always translate into health differences in the expected directions. Future research on duration patterns may need to consider alternative explanations beyond incorporation-based processes, such as cross-national health theories or age, period, cohort effects.3