Prior research has established associations between neighbourhood poverty and cumulative biological risk (CBR). CBR is conceptualized as indicative of the effects of stress on biological functioning, and is linked with increased morbidity and mortality. Studies suggest that supportive social relationships may be health protective, and may erode under conditions of poverty. This study examines whether social relationships are inversely associated with CBR and whether associations between neighbourhood poverty and CBR are mediated through social relationships. Data were from a stratified probability sample community survey (n=919) of residents of Detroit, Michigan, USA (2002–2003) and from the 2000 US Census. The outcome variable, CBR, included anthropometric and clinical measures. Independent variables included four indicators of social relationships: social support, neighbourhood satisfaction, social cohesion and neighbourhood participation. Multilevel models were used to test both research questions, with neighbourhood poverty and social relationships included at the block group level, and social relationships also included at the individual level, to disentangle individual from neighbourhood effects. Findings suggest some associations between social relationships and CBR after accounting for neighbourhood poverty and individual characteristics. In models that accounted for all indicators of social relationships, individual-level social support was associated with greater CBR (β=0.12, p=0.04), while neighbourhood-level social support was marginally significantly protective of CBR (within-neighbourhood: β=−0.36, p=0.06; between-neighbourhood: β=−0.24, p=0.06). In contrast, individual-level neighbourhood satisfaction was protective of CBR (β=−0.10, p=0.02), with no within-neighbourhood (β=0.06, p=0.54) or between-neighbourhood association (β=−0.04, p=0.38). Results indicate no significant association between either social cohesion or neighbourhood participation and CBR. Associations between neighbourhood poverty and CBR were not mediated by social relationships. These findings suggest that neighbourhood-level social support and individual-level neighbourhood satisfaction may be health protective and that neighbourhood poverty, social support and neighbourhood satisfaction are associated with CBR through independent pathways.