(1) Background: Social isolation and loneliness are determinants of healthy longevity. However, previous research has focused on either social isolation or loneliness and has not considered household types. This study sought to clarify loneliness and social isolation among older adults using single-person (ST) or multi-person (MT) household types. (2) Methods: We administered a national, anonymous, self-administered survey to 5351 Japanese older adults aged 65 years or older. The survey included subjects’ demographic characteristics and scores for loneliness (University of California Los Angeles (UCLA) Loneliness Scale version 3 (Cronbach’s α = 0.790)), social isolation (Lubben Social Network Scale (LSNS-6) (Cronbach’s α = 0.82)), and self-efficacy (GSES). (3) Results: After adjusting for age and gender, ST individuals had significantly lower LSNS-6 and significantly higher UCLA scores than MT individuals (p < 0.001). Lower LSNS-6 and higher UCLA scores were significantly associated with lower GSES scores, and the effect of GSES was greater for ST than for MT (LSNS-6, ST (β = 0.358, p < 0.001); MT (β = 0.295, p < 0.001)) (UCLA, ST (β = −0.476, p < 0.001); MT (β = −0.381, p < 0.001)). (4) Conclusions: Specific healthcare systems and programs based on self-efficacy should be developed by household type to reduce both social isolation and loneliness.