Background: There have been a limited number of quantitative studies on the relationship between coronary artery disease (CAD) and cardiorespiratory fitness (CRF), as measured by cardiopulmonary exercise testing (CPET). Thus, we aimed to investigate the association between CRF and the severity of coronary artery disease from the most comprehensive perspective possible, and to affirm the predictive value of CPET in the severity assessment of CAD. Methods: Our study included 280 patients with coronary angiography, who had undergone CPET in Tongji Hospital. The patients' CRF was measured through their peak oxygen uptake (VO2@peak), their oxygen uptake at the anaerobic threshold (VO2@AT) and through other parameters of CPET on a bicycle ergometer. The severity of the coronary artery disease was assessed in the following three layers: functionally significant lesions (quantitative flow ratio [QFR] <= 0.8), the number of stenotic coronary arteries (SCA, stenosis >= 50%) and the Gensini score. The correlation analyses were carried out between the CRF and the severity of the coronary artery disease. A ROC curve was plotted, and the AUC was calculated to distinguish the severe CAD and the non-severe CAD patients, as measured by the QFR, the number of SCA, and the Gensini score. Results: The VO2@AT and VO2@peak were inversely associated with the QFR. The VO2@AT, VO2@peak and VO2/kg@peak were associated with the number of SCA. Meanwhile, the VO2@AT, VO2/kg@AT, VO2@peak and VO2/kg@peak were associated with the Gensini score. An ROC analysis proved that a combination of traditional clinical risk factors and the VO2@peak/VO2prediction is valuable in predicting CAD severity. Conclusions: Our study demonstrated a strong and inverse association between CRF and the severity of CAD. A combination of traditional clinical risk factors and CRF is valuable in predicting CAD severity.