Precise relaxation time (RT) measurements have been made, with a standard deviation of 3% for T1 and T2 in white matter in normal volunteers. This sets an upper limit to the instrumental random errors (imprecision). Achieving this precision requires careful adjustment and use of the imager. The wide variation in RTs seen by other workers may be in part due to larger instrumental errors. We have measured RTs (both T1 and T2) in normal-appearing white matter in 16 normal controls and patients with multiple sclerosis (MS, 18), systemic lupus erythematosis (SLE, 16) and cerebral sarcoidosis (8). Both RTs were significantly higher in MS than in other patient groups and controls (P less than .05), possibly caused by microscopic lesions. T2 was elevated in SLE patients relative to controls and sarcoidosis patients (P less than .05), possibly because of microhemorrhages. Lesion RTs were abnormal but more variable and no significant differences between diseases were found.