Background <p>Mental health (MH) and substance use (SU) problems are prevalent among people in contact with the criminal justice system, many of whom serve all (or part) of their sentence in the community. Barriers to help-seeking in this population contribute to insufficient support, and later serious adverse outcomes for individuals and society. We aimed to understand variation in the relational context surrounding people in contact with the criminal justice system, and how these differences influence help-seeking for MH or SU problems.</p> Methods <p>We interviewed 50 people in contact with the criminal justice system in a mixed-methods social network analysis study. We collected data about participants’ social networks, help-seeking experiences, perceived societal stigma and perceived usefulness of MH and SU support. We developed a network typology by synthesising these data, and explored relationships between network types and help-seeking for MH or SU problems.</p> Results <p>We demonstrated the feasibility of our exploratory approach to developing a typology of social networks among people in contact with the criminal justice system in the community. Five network types were differentiated by structure, composition, network members’ characteristics, and culture. These could be applied to explore influences on a range of behaviours. Knowledge and attitudes towards MH, SU, and help-seeking across the network and within the wider community were important influences on individual help-seeking behaviour.</p> Conclusions <p>Interventions to encourage help-seeking, and other behaviours, among people in contact with the criminal justice system should consider network type and relational context. Introducing diverse contacts who can provide alternative perspectives may be useful for changing network culture towards different behaviours and facilitating individual change. Tailored messaging and delivery methods are required to increase awareness of MH and SU problems, and of when, where and how to seek help across the networks of people in contact with the criminal justice system. Further research could test our exploratory approach in larger samples and across contexts, examine the impact of network change over time, and rigorously develop and test relationally informed and context-responsive interventions.</p>