Background: Existing studies reporting on the levels of physical fitness among college students used relatively few fitness tests as a reflection of physical fitness, which could not comprehensively evaluate the levels of physical fitness. Thus, the current study aimed to investigate the cross-sectional relationship between body mass index (BMI) and a physical fitness index (PFI) based on six indicators of fitness in Chinese college students. Method: Anthropometric measurements and six measures of physical fitness (Vital capacity, 50-m sprint, sit and reach, standing long jump, 800/1000-m run, pull-up/bent-leg sit-up) were measured. BMI was calculated to classify individuals into underweight, normal weight, overweight, and obesity groups. Z-scores based on sex-specific mean and standard deviation were calculated, and the sum of z-scores for the six fitness tests was used as a PFI. Three models (a linear regression model, polynomial regression model with a second-order BMI term and a restricted cubic spline regression model) were fitted to discuss the potential relation between BMI and PFI. We compared the models using Akaike Information Criterion (AIC) and R square. Results: Totally, 8548 freshmen from the years 2014 to 2016 in a medical college completed the physical fitness tests. There was a decreasing trend of physical fitness index from the years 2014 to 2016 (P for trend < 0.01). More male than female students were overweight or obese (23.5% vs. 11.9%), but more female than male students were normal weight (74.7% vs. 64.8%). A restricted cubic spline regression model was superior to linear and polynomial regression model with lower AIC and higher R square. Conclusions: The relationships between BMI and PFI in college students were non-linear. Underweight, overweight and obese students had poorer performance in physical fitness index than normal weight students. Future prospective, longitudinal cohort studies to identify the causal relations and potential mechanism in a good manner are required.