Physical health is not only related to students’ personal health, but also directly affects the quality of education and the future development of the country. In this paper, a method for assessing college students’ physical fitness and health based on cluster analysis and logistic regression is proposed. First, the Relief algorithm is used to select features for students’ physical health data, and the improved K-means clustering algorithm is used to classify the data and analyze the physical characteristics of different classes. The results show that the improved K-means algorithm is significantly better than the original K-means algorithm in clustering effect, and the profile coefficient and Dunn’s index are 0.396658 and 0.043811, respectively, both of which are improved compared with the original algorithm. Then, the influencing factors of students’ physical health assessment were further analyzed based on the logistic regression model. The results showed that dynamic behavioral time, sleep duration and quality, and dietary and nutritional status had a significant effect on students’ physical health status, while static behavioral time and health knowledge did not significantly affect students’ physical health. The AUC values of the ROC curves of the model were 0.77, 0.87, and 0.83, respectively, indicating that the model has a good assessment performance. Eventually, a series of recommendations to improve the physical health of college students were proposed based on the model.
