The International Panel on the Diagnosis of Multiple Sclerosis first incorporated abnormalities demonstrated by brain and spinal cord magnetic resonance imaging (MRI) into the diagnostic criteria (McDonald criteria) for multiple sclerosis (MS), which were later revised in 2005. In 2006, Swanton and colleagues modified the MRI criteria to simplify and speed the diagnosis. The purpose of this study was to compare the ability of two sets of criteria (the revised McDonald MRI criteria and Swanton's modified criteria) to predict conversion from clinically isolated syndrome (CIS) to clinically definite multiple sclerosis (CDMS) from baseline MRI findings. Sixty-four patients presenting with CIS suggestive of multiple sclerosis were recruited from 2001 to 2006 and followed up for at least 2 years. Their baseline brain and spinal cord MRI studies were retrospectively evaluated. The patients who developed CDMS during follow-up were treated as positive cases. The sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), and accuracy of the two MRI dissemination-in-space criteria were calculated. Thirty patients (46.9%) converted to CDMS. The sensitivity specificity, PPV, NPV, and accuracy of the revised McDonald criteria (in %) were respectively 53, 100, 100, 71, and 78, while that of Swanton's modified criteria were 60, 100, 100, 74, and 81. In conclusion, Swanton's modified criteria are more sensitive and accurate (but not significantly so). However, Swanton's criteria are simpler to use and have equally high specificity and PPV.