ObjectiveTo determine the associations between social network size and subsequent long-term health behaviour patterns, as indicated by alcohol use, smoking, and physical activity. MethodsRepeat data from up to six surveys over a 15- or 20-year follow-up were drawn from the Finnish Public Sector study (Raisio-Turku cohort, n=986; Hospital cohort, n=7307), and the Health and Social Support study (n=20,115). Social network size was determined at baseline, and health risk behaviours were assessed using repeated data from baseline and follow-up. We pooled cohort-specific results from repeated-measures log-binomial regression with the generalized estimating equations (GEE) method using fixed-effects meta-analysis. ResultsParticipants with up to 10 members in their social network at baseline had an unhealthy risk factor profile throughout the follow-up. The pooled relative risks adjusted for age, gender, survey year, chronic conditions and education were 1.15 for heavy alcohol use (95% CI: 1.06–1.24), 1.19 for smoking (95% CI: 1.12–1.27), and 1.25 for low physical activity (95% CI: 1.21–1.29), as compared with those with >20 members in their social network. These associations appeared to be similar in subgroups stratified according to gender, age and education. ConclusionsSocial network size predicted persistent behaviour-related health risk patterns up to at least two decades.