A major aim in MS field has been the search for biomarkers that enable accurate detection of neuronal damage. Besides MRI, recent studies have shown that neuroaxonal damage can also be tracked by neurofilament detection. Nevertheless, before widespread implementation, a better understanding of the principal contributors for this biomarker is of paramount importance. Therefore, we analyzed neurofilament light chain (NfL) in relapsing (RMS) and progressive MS (PMS), addressing which MRI and clinical variables are better related to this biomarker. Forty-seven MS patients underwent MRI (3T) and cerebrospinal fluid (CSF) sampling. We measured NfL concentrations using ELISA (UmanDiagnostics) and performed multivariable regression analysis to assess the contribution of clinical and MRI metrics to NfL. NfL correlated with previous clinical activity in RMS (p < 0.001). In RMS, NfL also correlated with Gad+ and cortical lesion volumes. However, after multivariable analysis, only cortical lesions and relapses in previous 12 months remained in the final model (R2 = 0.610; p = 0.009 and p = 0.00008, respectively). In PMS, T1-hypointense lesion volume was the only predictor after multivariate analysis (R2 = 0.564; p = 0.012). CSF NfL levels are increased in RMS and associated with relapses and cortical lesions. Although NfL levels were correlated with Gad+ lesion volume, this association did not persist in multivariable analysis after controlling for previous clinical activity. We encourage controlling for previous clinical activity when testing the association of NfL with MRI. In PMS, the major contributor to NfL was T1-hypointense lesion volume. Copyright © 2019. Published by Elsevier B.V.