Cerebral blood flow (CBF) is a critical physiological process because it controls the oxygen supply, metabolic consumption, and byproduct clearance in the brain. However, a non-invasive method for long-term CBF monitoring is lacking. In recent years, NIR light has been used to monitor brain activities and cerebral blood flow based on DCS technique because it is able to penetrates human skull. The aim of this study was to develop a near infrared (NIR) Diffuse correlation spectroscopy (DCS) system for CBF monitoring. NIR laser at wavelength of 785 nm with properties of continuous wave and long coherence length (>10 m) was emitted into tissue. Single photons scattered by the red blood cells (RBCs) inside the cerebrovascular, were picked up. The auto-correlation function of the optical signal was calculated by the correlator downstream optical-detector. The developed NIR-DCS was first tested on a phantom in which the particle vibration was changed to simulate the changes in blood flow. Then the system was further tested on rats suffered with hypercapnia, normoxia and hyperoxia to measure the changes in CBF. The rats were connected to the ventilator through two plastic tubes—one for inhalation, the other for exhalation. The content of the exhaled gas was analyzed, and the real-time partial pressure of CO2 and the current end tidal CO2 (EtCO2) are measured. The results showed that hyperoxia increased blood flow due to changes in vascular wall tension. The outcome from this study supported DCS as a novel noninvasive method to measure CBF.