The first aim of this study was to investigate retest-reliability of gray matter perfusion estimates in patients with multiple sclerosis. Using multi-inversion time pulsed ASL, we demonstrate good repeatability of global and local perfusion estimates over a four-week interval. The perfusion estimates were consistently lower in the patients than in a group of healthy controls. The second aim was to relate GM perfusion to disease characteristics. We could not find an effect of disease duration or stage (relapsing-remitting vs progressive) on perfusion, but we did observe a weak correlation with cognition, supporting previous studies.