Changes on serial assessments of brain MRI lesion load are used for monitoring therapeutic efficacy in patients with multiple sclerosis (MS). We assessed the accuracy and reliability of conventional spin-echo (CSE) and fast spin-echo (FSE) sequences for measurement of lesion volume using a semiautomated contour technique. Cranial CSE and FSE examinations of 18 patients with secondary progressive MS were studied. The mean lesion load was slightly higher with the CSE sequence (p = 0.002). Intraobserver variability was significantly higher for FSE than for CSE, according to both the coefficient of variation between two measurements (mean 2.48% and 1.35% respectively, p < 0.05) and back-transformed 95% limits of agreement (1.005-1.060 for FSE; 0.988-1.019 for CSE). Although FSE sequences are quicker and the total lesion volume measurements are similar to those obtained with CSE, the poorer reproducibility raises doubts about the use of FSE to replace CSE in clinical trials.