**Background:** First-semester university students are at high risk for adopting health-risk behaviors including unhealthy eating, physical inactivity, and problematic alcohol use. In Colombia, high rates of overweight/obesity and weekly alcohol consumption have been reported among university students. The health behaviors of close social ties (parents, partners, peers), co-residence with these ties, and time spent with peers may influence students' current and expected health behaviors. However, limited research has examined these factors simultaneously in Latin American populations.
**Methods:** A cross-sectional quantitative design was used. Participants were 189 first-semester Colombian university students (M_age = 18.79; SD = 1.07; 68.8% female; 66.7% lived with parents; 32.3% had a partner; 95.2% from Colombia). Data were collected via online questionnaire from end of July to end of August 2022. Measures included: unhealthy food consumption (5 items adapted from CDC's NCHRBS; α = 0.72 current, 0.87 expected), physical activity (4 items from NCHRBS; α = 0.80 current, 0.90 expected), alcohol use (3 items from AUDIT assessing frequency, quantity, and binge drinking), perceived health behaviors of social ties (9 items), and social ties' encouragement of risk behaviors (9 items). ANCOVAs, bivariate correlations, moderation analyses (multiple regression), and hierarchical regressions were used.
**Key Results:** Hypothesis 1 (students who moved out expect stronger increases in unhealthy behaviors) was not supported. However, students who moved out already showed significantly less physical activity (M = 23.37 vs. 21.17; p = 0.04) and more binge drinking (M = 2.13 vs. 1.77; p = 0.02) than those who did not move out. Hypothesis 2 (health behaviors correlated with social ties' behaviors) was partially supported: expected food consumption correlated with parental food consumption (r = 0.18, p = 0.01); current and expected heavy drinking (r = 0.22, p = 0.01; r = 0.23, p = 0.01) and binge drinking (r = 0.34, p = 0.001; r = 0.27, p = 0.001) correlated with parental alcohol use; partners' alcohol use correlated with participants' current heavy drinking (r = 0.34, p = 0.01) and current/expected binge drinking (r = 0.54, p = 0.001; r = 0.34, p = 0.01); peers' physical activity correlated with expected physical activity (r = 0.20, p = 0.01). Hypothesis 3 (stronger correlations with co-residence) was supported only for expected food consumption with parents (stronger among those living with parents: β = 0.96, p = 0.04 vs. β = 0.75, p = 0.13). For alcohol, associations were stronger among those not living with parents (β's 0.38–0.60, p's 0.01–0.001 vs. β's 0.08–0.14, p's 0.11–0.36). Hypothesis 4 (social ties' encouragement moderates associations) was supported only for partners' encouragement of drinking, which strengthened the association between participants' binge drinking and partners' alcohol use (β's 0.30–0.40, p's 0.01–0.02 for high encouragement vs. β's 0.05–0.10, p's 0.51–0.92 for low encouragement). Hypothesis 5 (more time with peers → unhealthier behaviors) was partially supported: more time with peers was associated with more drinks per month (significant in hierarchical regression) and with more physical activity, but the binge drinking association did not reach significance.
**Clinical Implications:** In Colombia, parents remain influential on their offspring's food consumption and alcohol use during the first university semester, suggesting that family-based interventions may be effective. Partner drinking, particularly binge drinking reinforced by partner encouragement, represents a key target for intervention. Time spent with peers is a double-edged sword—associated with both heavier drinking and more physical activity. Interventions should educate parents about modeling healthy behaviors, limit alcohol availability at home, and address partner influences on drinking. Parental monitoring and limiting unhealthy peer activities should be encouraged, especially in collectivistic cultures where parental influence is stronger. Limitations include the exclusion of students under 18, insufficient power to analyze co-residence with partners/peers, and the cross-sectional design preventing causal inference.