Objective: To explore the patterns of health risk behaviors among secondary school students in Shanghai, so as to provide a basis for implementing targeted interventions. Methods: From May to June 2021, 21 junior high schools, regular high schools and secondary vocational schools were selected from Shanghai, and the occurrence of dangerous health behaviors of 6 092 students were investigated in the form of self-filled questionnaires. The clustering analysis of the sample was performed by Ward s departure and peace method. Results: Three types of clustering characteristics were found in this study: being bullied, depression (64.9%), loneliness (48.2%), and insomnia (42.4%) in the high-risk group (1 004, 16.5%); excessive video games (53.1%) and sugary drinks (89.3%) and fried food intake (79.0%) in the medium-risk group (1 585, 26.0%); the low-risk group (3 503, 57.5%) was characterized by extremely low alcohol consumption (0) and smoking (0.4%). The reporting rate on sugary beverage intake, fried food intake, physical activity, malicious teased, solicited, isolated/intentionally excluded, outside the group activities, threatened, kicked/hit/pushed/locked in house, smoking, unhealthy weight loss, playing video game overtime were statistically significant among the three groups ( χ 2=36.74, 41.85, 30.07, 652.68, 334.40, 669.89, 358.84, 233.12, 101.46, 88.45, 230.15, P <0.01) .The difference in reporting rates of loneliness, depression, insomnia and drinking were significant among the three groups (Fisher s test, P <0.01). The majority of junior high school students, general high school students and girls in the high-risk group; boys, general high school students and junior high school students in the excessive video game and unhealthy diet group. Conclusion There are clusters of health-related risk behaviors among secondary school students in Shanghai, focusing on two major aspects: poor lifestyle and mental health. Targeted interventions should be implemented according to the characteristics of health-related risk behaviors clustering to enhance the intervention effect.