Functional connectivity (FC) in fMRI has generated major attention recently. Patterns of FC are consistently found in healthy subjects, whereas alterations of these patterns have been associated with many neuropsychiatric disorders. However, confounding factors arising from physiological processes have to be taken into consideration when analyzing and interpreting the results in order to ensure their validity. Even though physiological noise correction is commonly applied to fMRI, it is believed that the field would certainly benefit from more efficient techniques. In this study, we examine the relationship of the global BOLD signal with fluctuations in heart rate and respiration and propose a new method for removing the associated artifacts from whole-brain fMRI data.