Magnetic resonance imaging (MRI) may demonstrate disease activity in a number of ways in patients with multiple sclerosis. Newly appearing MRI lesions, gadolinium-enhancing lesions, and enlargement of preexisting lesions are frequently taken as evidence of disease activity. Furthermore, serial MRI studies have been stated to be more sensitive than repeated neurologic examinations in detecting disease activity. We assessed the validity of using lesion enlargement as a measure of disease activity by repeatedly measuring the area of all MRI lesions in four patients with multiple sclerosis. The size-frequency distribution in all patients was similar, with 80% of the lesions measuring less than 0.67 cm2. The median coefficient of variation for three successive lesion measurements was inversely related to lesion area, ranging from 22.6% for the more common smaller lesions (less than 0.67 cm2) to 12.1% for larger lesions. Based on these coefficients of variation, a change in a particular lesion exceeding 45.2% for a baseline lesion smaller than 0.67 cm2 and 24.2% for a baseline lesion greater than or equal to 0.67 cm2 should be required to exclude a change due to measurement variability. It remains necessary to determine the number of lesions that must change when multiple lesions are present in the baseline MRI to reliably exclude chance occurrence when establishing MRI-evident disease activity. Guidelines for determining these criteria are presented, as are the limitations inherent in the statistical model employed to make these determinations.