It has been shown that serum brain-derived neurotrophic factor (BDNF) is associated with skeletal muscle energy metabolism and that BDNF is a predictor of mortality in heart failure patients. However, little is known about the relationship between BDNF and cardiac rehabilitation (CR). Therefore, this study retrospectively investigated the effects of baseline serum BDNF levels on the CR-induced exercise capacity improvement in patients with cardiovascular disease (CVD). We assigned 99 CVD patients (mean age 71±12 years, male = 60) to Low, Middle, and High groups based on the tertiles of baseline BDNF levels. Cardiopulmonary exercise testing was done using supervised bicycle ergometer twice before and after 3 weeks of CR. Analysis of covariance (ANCOVA) followed by post-hoc analysis using Tukey's HSD test was conducted to assess the multivariate associations between baseline BDNF levels categorized by BDNF tertiles (as independent variable) and %increases in AT and peak VO2 after 3-week CR (as dependent variables) after adjustment for age and gender (as covariates), as a main statistical analysis of the present study. The higher the baseline BDNF levels, the better nutritional status evaluated by the CONUT score (p<0.0001). Baseline anaerobic threshold (AT) and peak oxygen uptake (peak VO2) were similar among the three groups. ANCOVA followed by post-hoc analysis revealed that age- and gender-adjusted %increases in peak VO2 after 3-week CR were positively associated with baseline BDNF levels (p = 0.0239) and Low BDNF group showed significantly lower %increase in peak VO2 than High BDNF group (p = 0.0197). Significant association was not found between baseline BDNF and %increase in AT (p = 0.1379). Low baseline BDNF levels were associated with malnutrition in CVD patients. A positive association between baseline BDNF levels and CR-induced increases in peak VO2 was found. It was suggested that CVD patients with low baseline BDNF levels may be poor responders to CR.