Central neuropathic pain (CNP) occurs in many multiple sclerosis (MS) patients. The provision of adequate pain relief to these patients can very difficult. Here we report the first phase III placebo-controlled study of the efficacy of the endocannabinoid system modulator delta-9-tetrahydrocannabinol (THC)/cannabidiol (CBD) oromucosal spray (USAN name, nabiximols; Sativex, GW Pharmaceuticals, Salisbury, Wiltshire, UK), to alleviate CNP. Patients who had failed to gain adequate analgesia from existing medication were treated with THC/CBD spray or placebo as an add-on treatment, in a double-blind manner, for 14 weeks to investigate the efficacy of the medication in MS-induced neuropathic pain. This parallel-group phase of the study was then followed by an 18-week randomized-withdrawal study (14-week open-label treatment period plus a double-blind 4-week randomized-withdrawal phase) to investigate time to treatment failure and show maintenance of efficacy. A total of 339 patients were randomized to phase A (167 received THC/CBD spray and 172 received placebo). Of those who completed phase A, 58 entered the randomized-withdrawal phase. The primary endpoint of responder analysis at the 30 % level at week 14 of phase A of the study was not met, with 50 % of patients on THC/CBD spray classed as responders at the 30 % level compared to 45 % of patients on placebo (p = 0.234). However, an interim analysis at week 10 showed a statistically significant treatment difference in favor of THC/CBD spray at this time point (p = 0.046). During the randomized-withdrawal phase, the primary endpoint of time to treatment failure was statistically significant in favor of THC/CBD spray, with 57 % of patients receiving placebo failing treatment versus 24 % of patients from the THC/CBD spray group (p = 0.04). The mean change from baseline in Pain Numerical Rating Scale (NRS) (p = 0.028) and sleep quality NRS (p = 0.015) scores, both secondary endpoints in phase B, were also statistically significant compared to placebo, with estimated treatment differences of -0.79 and 0.99 points, respectively, in favor of THC/CBD spray treatment. The results of the current investigation were equivocal, with conflicting findings in the two phases of the study. While there were a large proportion of responders to THC/CBD spray treatment during the phase A double-blind period, the primary endpoint was not met due to a similarly large number of placebo responders. In contrast, there was a marked effect in phase B of the study, with an increased time to treatment failure in the THC/CBD spray group compared to placebo. These findings suggest that further studies are required to explore the full potential of THC/CBD spray in these patients.