AIMS/INTRODUCTION: The principal aim of the present study was to investigate the cardiovascular autonomic system status of diabetes patients using approximate entropy (ApEn) extracted from 24-h heart rate variability (HRV) and its frequency components. MATERIALS AND METHODS: A total of 29 healthy controls and 63 type 2 diabetes patients were included. Participants\' 24-h HRV signals were recorded, and decomposed and reconstructed into four frequency components: high, low, very low and ultra low. The total 24-h HRV and its four components were divided into 24 1-h segments. ApEn values were extracted and statistically analyzed. Four traditional HRV indices, namely standard deviation of the RR intervals, root mean square of successive differences, coefficient of variance of RR intervals and ratio of low to high power of HRV, were also calculated. RESULTS: The low-frequency component contained the most abundant non-linear information, so was potentially most suitable for studying the cardiovascular system status with non-linear methods. ApEn values extracted from low- and high-frequency components of healthy controls were higher than those of diabetes patients. Except for root mean square of successive differences, standard deviation of the RR intervals, low to high power of HRV and coefficient of variance of RR intervals of healthy controls were all higher than those of diabetes patients. CONCLUSIONS: The results showed that ApEn contained information on disorders of autonomic system function of diabetes patients as traditional HRV indices in time and frequency domains. ApEn and three traditional indices showed accordance to some degree. Non-linear information in subcomponents of HRV was shown, which is potentially more effective for distinguishing healthy individuals and diabetes patients than that extracted from the total HRV. Compared with diabetes patients, the cardiovascular system of healthy controls showed information of higher complexity, and better regulation function in response to changes of environment.