Few attempts have been made to detect subtle blood-brain barrier (BBB) leakage in visibly non-enhancing MRI lesions in multiple sclerosis (MS). For 19 patients, longitudinal relaxation time (T1) maps were generated from MRI scans obtained before, and at 20, 40 and 60 minutes after injection of gadolinium (Gd)-DTPA (0.3 mmol/kg). Regions of interest (ROI) were placed around non-enhancing lesions, and in paired contralateral normal appearing brain tissue (NABT). Post-Gd rate of R1 (=1/T1) rise (DeltaR1/Deltat), was used to quantify leakage. DeltaR1/Deltat was greater in lesions than paired NABT (P < or = 0.001 at all post-Gd timepoints). DeltaR1/Deltat was greater in T1 hypointense than isointense lesions (P = 0.001 and 0.01 for first and second timepoints respectively), and negatively related to lesion cross sectional area (P < or = 0.001 at all post-Gd timepoints). Relapsing remitting (RRMS) lesions had a greater initial DeltaR1/Deltat than secondary progressive (SPMS) lesions ( P = 0.04), but this was not seen in subsequent timepoints. DeltaR1/Deltat in visibly enhancing lesions was significantly greater than in visibly non-enhancing lesions, with no overlap in the normal ranges of the two populations. Subtle BBB leakage is a consistent feature in non-enhancing lesions, and is distinct from the overt BBB leakage observed in visibly enhancing lesions. It is detectable using quantitative contrast-enhanced MRI. It is apparent in all clinical and lesion subtypes studied, and greater in T1 hypointense and smaller lesions. Larger initial DeltaR1/Deltat in RRMS than SPMS lesions may reflect differences in blood volume rather than BBB leakage.