Spinal and bulbar muscular atrophy is an X-linked neuromuscular disease caused by CAG repeat expansion in the androgen receptor gene. Patients with this disease have low concentrations of insulin-like growth factor-1 (IGF-1), and studies of overexpression and administration of IGF-1 showed benefit in a transgenic model; thus the IGF-1 pathway presents as a potential treatment target. We assessed safety, tolerability, and preliminary efficacy of BVS857, an IGF-1 mimetic, in patients with spinal and bulbar muscular atrophy. In this randomised, double-blind, placebo-controlled trial, we recruited patients from neuromuscular centres in Denmark (Copenhagen), Germany (Ulm), Italy (Padova), and three sites within the USA (Bethesda, MD; Irvine, CA; and Columbus, OH). Eligible patients were 18 years or older with a confirmed genetic diagnosis of spinal and bulbar muscular atrophy, were ambulatory, had symptomatic weakness, and had serum IGF-1 concentrations of 170 ng/mL or lower. Patients were randomly assigned (2:1) to study drug or placebo by a number scheme. Patients, investigators, and study personnel were masked to treatment assignment. After a safety and tolerability assessment with eight patients, BVS857 was administered once a week (0·06 mg/kg intravenously) for 12 weeks. Primary outcome measures were safety, tolerability, and the effects of BVS857 on thigh muscle volume (TMV) measured by MRI. The ratio of TMV at day 85 to baseline was analysed with ANCOVA per protocol. Secondary outcomes of muscle strength and function were measured with the Adult Myopathy Assessment Tool, lean body mass through dual energy x-ray absorptiometry, and BVS857 pharmacokinetics. This trial was registered with ClinicalTrials.gov, NCT02024932. 31 patients were assessed for eligibility, 27 of whom were randomly assigned to either BVS857 treatment (n=18) or placebo (n=9), and 24 were included in the preliminary efficacy analysis (BVS857 group, n=15; placebo group, n=9). BVS857 was generally safe with no serious adverse events. No significant differences were found in adverse events between the BVS857 and placebo groups. Immunogenicity was detected in 13 (72%) of 18 patients in the BVS857 group, including crossreacting antibodies with neutralising capacity to endogenous IGF-1 in five patients. TMV decreased from baseline to day 85 in the placebo group (-3·4% [-110 cm3]) but not in the BVS857 group (0% [2 cm3]). A significant difference in change in TMV was observed in the BVS857 group versus the placebo group (geometric-mean ratio 1·04 [90% CI 1·01-1·07]; p=0·02). There were no differences between groups in measures of muscle strength and function. TMV remained stable in patients with spinal and bulbar muscular atrophy after being given BVS857 for 12 weeks. The intervention was associated with high incidence of immunogenicity and did not improve muscle strength or function. Additional studies might be needed to assess the efficacy of activating the IGF-1 pathway in this disease. Novartis Pharmaceuticals and the US National Institutes of Health.