Although social engagement is important to health behaviors, prior studies mostly considered only one of multiple different dimensions of it, with inconsistent results. Presence of countervailing or interacting influences of different types of social engagement may account for prior conflicting findings. This study identifies which dimensions of social engagement are associated with risk behaviors of smoking, low levels of physical activity, and a diet poor of fruit and vegetable intake. Cross-sectional data from 5004 participants, aged 45-84 from the Multi-Ethnic Study of Atherosclerosis was used. Social support was measured using the ENRICHD Social Support Instrument, loneliness with a three item instrument derived from the revised University of California at Los Angeles Loneliness Scale, and neighborhood social cohesion with the instrument from the Project on Human Development in Chicago Neighborhoods. Prevalence ratios of behaviors associated with standardized social engagement variables were modeled with Poisson regression, using robust standard errors. In unadjusted analyses, higher social support and neighborhood social cohesion were associated with slightly lower prevalence of smoking (prevalence ratio, PR=0.82, 95% CI: 0.77-0.87, and PR=0.91, 95% CI: 0.84-0.98, respectively) and slightly higher likelihood of achieving recommended levels of physical activity (PR=1.04, 95% CI: 1.02-1.07, and PR=1.05, 95% CI: 1.03-1.07, respectively). Higher loneliness was associated with slightly higher prevalence of smoking (PR=1.17, 95% CI: 1.10-1.24) and slightly less physical activity (PR=0.95, 95% CI: 0.93-0.98 ) . Associations of social support and loneliness with smoking, and associations of social support and neighborhood cohesion with physical activity, persisted after adjusting for individual characteristics. After further adjusting for all dimensions of social engagement, only neighborhood social cohesion remained associated with very slightly higher probability of achieving recommended physical activity levels (PR=1.03, 95% CI: 1.01, 1.05), and only social support remained associated with lower prevalence of smoking (PR=0.87, 95% CI: 0.81, 0.94) and higher probability of quitting, amongst those who had ever smoked (PR=1.03, 95% CI: 1.01, 1.06). Social support interacted with loneliness (p for interaction=0.001), such that these associations were present only amongst non-lonely individuals (PR=0.77, 95% CI: 0.69, 0.86). In fully adjusted models, no associations were found for fruit and vegetable intake. This study is among the first to simultaneously investigate several dimensions of social engagement and their interactions in relation to risk behaviors. Social support is more consistently associated with smoking than other dimensions of social engagement and this association appears to be modified by loneliness.