This study used an infusion of a paramagnetic contrast agent to perturb intravascular blood susceptibility and investigate its effect on the BOLD hemodynamic response. A three compartment BOLD signal model combined with a modified balloon model was developed to interpret the MR signal. This model incorporated arterial blood volume in order to simulate signal changes resulting from the contrast agent. The BOLD signal model was fitted to the experimental data to test the hypothesis that arterial blood volume changes during activation. It was found that allowing arterial blood volume to change, rather than assuming this change is negligible as often assumed in the literature, provides a better fit to the experimental data, particularly during the BOLD overshoot. The post-stimulus undershoot was fitted well, regardless of whether the arterial blood volume was allowed to change, by assuming that this feature is due to delayed venous compliance. However the resultant elevation in post-stimulus blood volume decays with an extremely long time constant, taking more than 55 s to recover to baseline following a 4.8 s stimulus. The post-stimulus signal changes measured here could alternatively be described by a post-stimulus elevation in metabolism. An alternative model of oxygen extraction, in place of the Oxygen Limitation model, would be required to test this hypothesis.