Accurate cerebral blood flow (CBF) quantification using arterial spin labelling (ASL) relies on physiological and MR parameters. Longitudinal relaxation time (T1) of blood, which depends on haematocrit, can be a factor in some patient groups. We determined subject-specific T1 using the DESPOT-1 HIFI method in a mild stroke cohort, calculating CBF using nominal and subject-specific values. CBF calculated with subject-specific T1 values was lower in grey and higher in white matter, though there was not a proportional bias. CBF was lower in patients with higher disease burden. Subject-specific T1 values can reduce variance, potentially improving CBF quantification in clinical ASL.