Background: Muscle spasticity is common in multiple sclerosis (MS), occurring in more than 60\% of patients. Objective: To compare Sativex with placebo in relieving symptoms of spasticity due to MS. Methods: A 15-week, multicenter, double-blind, randomized, placebo-controlled, parallel-group study in 337 subjects with MS spasticity not fully relieved with current anti-spasticity therapy. Results: The primary endpoint was a spasticity 0-10 numeric rating scale (NRS). Intention-to-treat (ITT) analysis showed a non-significant improvement in NRS score, in favor of Sativex. The per protocol (PP) population (79\% of subjects) change in NRS score and responder analyses (>=30\% improvement from baseline) were both significantly superior for Sativex, compared with placebo: -1.3 versus -0.8 points (change from baseline, p=0.035); and 36\% versus 24\% (responders, p=0.040). These were supported by the time to response (ITT: p=0.068; PP: p=0.025) analyses, carer global impression of change assessment (p=0.013) and timed 10-meter walk (p=0.042). Among the subjects who achieved a >=30\% response in spasticity with Sativex, 98, 94 and 73\% reported improvements of 10, 20 and 30\%, respectively, at least once during the first 4 weeks of treatment. Sativex was generally well tolerated, with most adverse events reported being mild-to-moderate in severity. Discussion and conclusions: The 0-10 NRS and responder PP analyses demonstrated that Sativex treatment resulted in a significant reduction in treatment-resistant spasticity, in subjects with advanced MS and severe spasticity. The response observed within the first 4 weeks of treatment appears to be a useful aid to prediction of responder/non-responder status.