Objectives We sought to characterize the social networks of older adults who report feeling lonely or like a burden on others, psychological states that are associated with risk for suicide according to the Interpersonal Theory of Suicide. Methods We used a latent class analysis to identify distinct groups of older adults based on social network characteristics and perceptions of their networks within a sample of older adults endorsing loneliness and/or feeling like a burden. We examined associations between class membership and mental health outcomes. Results Four network types were identified: small, cohabitating networks with daily contact; moderate-sized family-oriented networks with multiple contacts weekly; moderate-sized friend-oriented networks with weekly contact; and average-sized mixed networks with weekly contact. The friend-oriented class reported the greatest loneliness, perceived burden, and lifetime prevalence of suicidal ideation and attempts. Conclusions Social network composition may be more explanatory of loneliness and perceived burden than number of members alone. Profile differences in outcomes suggest utilizing tailored social connectedness interventions. Individuals with small-to-moderate networks may benefit most from interventions designed to build friendships. Individuals with many close confidants may benefit most from psychotherapy. Individuals reporting moderately frequent contact with moderately-sized networks may benefit from both intervention types.