Personal networks significantly contributed to initiating risky behaviours among emerging youth. The study aimed to assess the personal network characteristics and factors associated with the personal network size of underage alcohol users. 200 underage drinkers were recruited through respondent-driven sampling in Dibrugarh, Assam using five seeds. Transition probability matrices and homophily indices were used to describe underage drinkers’ network characteristics and network patterns. The weighted logistic regression model was performed to observe the association. Personal network size differed significantly with respect to demographic profile and alcohol consumption patterns of the respondents. A higher proportion of participants used cannabis (77.0%), drank for more than 3 years (40.7%) and liked to drink with peers (51.1%). The affiliation pattern of educational status shows a trend of homophily and interaction between new and older underage drinkers. Students (OR 3.17, 95% CI 1.32, 7.62) who like to drink with peers (OR 3.45, 95% CI 1.24, 9.62) were found significantly associated with larger personal network size in weighted multiple binary logistic regression. The findings will be helpful to understand the network pattern of underage drinkers. The study highlighted the need for effective prevention and intervention programmes for this population and suggested further research in this area.