Abstract: Objective This study aimed to investigate the relationship between homeostatic model assessment of insulin resistance (HOMA-IR), body composition and physical exercise of adolescents. Methods A total of 264 students aged 10-16 from No. 7 middle school (ordinary middle school group) and Jining Sports Training Center (sports school group) were randomly selected as the research subjects. During the baseline period, height, weight, blood pressure, heart rate, waist circumference, hip circumference and other data were collected through physical examination. Fasting blood glucose (FBG) and insulin (INS) levels were measured, self-reported dietary information (frequency of sweets, carbonated drinks, and milk intake), time spent using electronic devices, and nighttime sleep time were collected through questionnaires. During the follow-up period after 1 year, the data of body composition weight, skeletal muscle mass, body fat mass and body fat percentage were measured. The correlation between HOMA-IR and body composition was analyzed by single factor linear regression model. The Regression analysis of long-term physical activity and physical activity in general secondary schools of both groups were taken as the dependent variable, and the group was taken as the independent variable, age, sex at follow-up and data at baseline were included as covariates in the model. Results A total of 264 students participated in the data collection in the baseline period and the follow-up period. Among them, 154 students in the ordinary middle school group participated in physical exercise activities inside and outside the school freely; 110 students in the physical education school group, strictly participate in different types of school physical training, from the baseline period to follow-up period to maintain regular physical exercise. The body mass index (BMI), INS level and HOMA-IR of the students in the sports school group were lower than those of the students in the ordinary middle school group at the baseline and follow-up periods (P < 0.005). HOMA-IR was positively correlated with BMI, waist circumference and body fat rate in both groups. Models of HOMA-IR, INS level, BMI, waist circumference, and body fat percentage were used to evaluate the effects of age, gender, and body fat percentage on the outcome of the two groups of students, and self-reported diet (frequency of sweets, carbonated drinks, and milk intake), screen time, and night-time sleep, adjusted for body fat percentage, the remaining models also need to be adjusted for baseline dependent variables. The results showed that long-term regular physical exercise was negatively correlated with HOMA-IR, INS level, BMI, waist circumference and body fat rate during follow-up period. Students in the general secondary school group were further divided into a physical activity subgroup of 27 and a non-physical activity subgroup of 127 based on self-reported physical activity in questionnaires during the baseline and follow-up periods, after adjusting the influencing factors, it was found that physical exercise was also negatively correlated with HOMA-IR, INS level, BMI, waist circumference and body fat rate during the follow-up period. Conclusions Long-term regular physical exercise can improve BMI and insulin sensitivity in adolescents, which may help to prevent and improve excessive fat accumulation and insulin resistance.