The interobserver variation in the magnetic resonance (MR) location of cerebral vein and dural sinus thrombosis (CVT) has not been previously reported. Four independent observers rated a convenience sample of 40 MR/MR angiographies to assess whether or not each dural sinus and major cerebral veins were occluded. Interobserver reliability was measured using kappa statistics. Interobserver agreement was comparable between the six pairs of raters. Agreement was excellent for thrombosis of the deep cerebral venous system (kappa = 1.00), cerebellar veins (kappa = 1.00), superior sagittal sinus (kappa range: 0.82-1) and right jugular vein (kappa range: 0.84-0.95); good to excellent for the right transverse/sigmoid sinus (kappa range: 0.75-0.90) and the left jugular vein (kappa range: 0.65-0.85); moderate to excellent for the left lateral sinus (kappa range: 0.59-0.78) and the straight sinus (kappa range: 0.59-0.92); poor to good for the cortical veins (kappa range: 0.02-0.65). Agreement between observers varies with the location of CVT. It is good or excellent for most of the occluded sinus and veins, except for the cortical veins. This study suggests that information on the location of CVT can be reliably collected and used in multicentre studies.