Caffeine is the most widely consumed methylxanthine, primarily found in beverages. As a nonspecific antagonist to adenosine, caffeine binds to both A₁ and A_2A adenosine receptors to increase neural activity and decrease cerebral blood flow, which allows it to alter neurovascular coupling. However, there has been much debate about the effects of caffeine on functional activity measured by BOLD imaging because of the complexity of the BOLD signal. Arterial spin labeling, on the other hand, is capable of measuring blood flow directly and quantitatively. A recently proposed calibrated BOLD model that combines ASL and BOLD with CO₂-induced hypercapnia is an excellent tool for studying the effect of substances such as caffeine on functional activity.

The work in this dissertation is divided into two parts. The first part consists of studies performed to optimize the parameters for a calibrated BOLD study of caffeine. The standard ASL sequence was first modified to streamline the CBF quantification process by employing an entirely automatic algorithm. This algorithm was used to investigate the reproducibility of pulsed ASL in healthy subjects. Clinical feasibility of this algorithm was assessed in stroke subjects. Then, the sensitivity of simultaneous ASL/BOLD imaging was investigated for simple visual and motor tasks. This simultaneous method was then used to determine the dose-response of caffeine's effects on task-induced brain activation. A "trace" method that monitors the effects of caffeine during injection showed that caffeine exerts its actions within a minute after the onset of injection and visibly reduces physiological fluctuations in the brain.

The second part of this thesis focuses on the calibrated BOLD approach. CBF:CMRO₂ coupling ratio was determined in motor and visual cortices before and after a dose of 2.5mg/kg caffeine. The results demonstrated that caffeine decreases CBF:CMRO₂ coupling, possibly through a combination of increased oxygen extraction fraction and glycolysis.