Introduction: Dimethyl fumarate treatment is approved in Europe for patients with relapsing-remitting multiple sclerosis (MS) and in the US for relapsing forms of MS. We recently published the results of the first randomized placebo-controlled trial of 48 weeks of treatment with dimethyl fumarate or placebo in primary progressive MS (PPMS) (clinicaltrial.gov NCT02959658). The placebo-controlled phase of the trial did not meet its primary endpoint (reduction in cerebrospinal fluid concentrations of neurofilament light chain [NFL]).Aim: To investigate the effects of dimethyl fumarate treatment in the open-label extension phase of the trial (week 48-96), where all patients were treated with DMF.Methods: Reported data are from screening, week 48, and week 96 visits. Patients were clinically evaluated with Expanded Disability Status Scale (EDSS), 9-Hole Peg Test (9HPT), Timed 25-Foot Walk (T25FW) test, Symbol Digit Modalities Test (SDMT), California Verbal Learning Test, and Brief Visuospatial Memory-Revised. Serum NFL concentrations were measured by single-molecule array analysis. MRI was performed on a 3 tesla MRI scanner and included: new/enlarging lesions, volume of lesions, cortical grey matter, putamen, thalamus, and normal-appearing white matter, and additional diffusion tensor imaging and magnetization transfer ratio measures.Results: Forty-two patients entered the open-label treatment phase, and 33 patients (61%) had complete data sets at week 96. The remaining 39% did not complete the trial and were not evaluated at week 96. We found no evidence of differences in clinical and MRI measures between patients initially treated with dimethyl fumarate and patients initially treated with placebo from baseline to week 48 and from week 48-96, where all patients were treated with dimethyl fumarate. Serum NFL concentrations remained stable in both groups over 96 weeks. Assessed with either EDSS, T25FW, or 9HPT at week 96, progression was observed for 14 patients (45%). Interestingly, another 15 patients (46%) had improvement in one or more of these domains. Applying a cut-off of 8 points, 2 (6%) patients worsened on SDMT, 25 (78%) did not change, and 5 (16%) improved.Conclusions: Dimethyl fumarate treatment showed no effects on neither clinical nor MRI outcomes or changes in serum concentrations of NFL. An expected number of patients showed evidence of progression on standard clinical scales; however, this was matched by an equal number of patients improving. The reasons for the physical improvement in an unexpectedly high proportion of patients must be addressed in future studies.