BACKGROUND: Myostatin inhibits skeletal muscle growth. The humanised monoclonal antibody LY2495655 (LY) binds and neutralises myostatin. We aimed to test whether LY increases appendicular lean body mass (aLBM) and improves physical performance in older individuals who have had recent falls and low muscle strength and power. METHODS: In this proof-of-concept, randomised, placebo-controlled, double-blind, parallel, multicentre, phase 2 study, we recruited patients aged 75 years or older who had fallen in the past year from 21 investigator sites across Argentina, Australia, France, Germany, Sweden, and the USA. Eligible patients had low performance on hand grip strength and chair rise tests, tested with the procedure described by Guralnik and colleagues. Participants were stratified by country, age, hand grip strength, and performance on the chair rise test, and were randomly assigned (1:1) by a computer-generated random sequence to receive subcutaneous injections of placebo or 315 mg LY at weeks 0 (randomisation visit), 4, 8, 12, 16, and 20, followed by 16 weeks observation. The primary outcome was change in aLBM from baseline to 24 weeks. We measured physical performance as secondary outcomes (four-step stair climbing time, usual gait speed, and time to rise five times from a chair without arms, or with arms for participants unable to do it without arms) and exploratory outcomes (12-step stair climbing test, 6-min walking distance, fast gait speed, hand grip strength, and isometric leg extension strength). Efficacy analyses included all randomly assigned patients who received at least one dose and had a baseline and at least one subsequent measure. The primary analysis and all other tests of treatment effect (except physical performance tests) were done at a two-sided alpha level of 0·05. Tests of treatment effect on physical performance tests were done at a pre-specified two-sided alpha level of 0·1. This trial is registered with ClinicalTrials.gov, number NCT01604408. FINDINGS: Between June 19, 2012, and Dec 12, 2013, we screened 365 patients. 99 were randomly assigned to receive placebo and 102 to receive LY. Treatment was completed in 85 (86%) of patients given placebo and in 82 (80%) given LY. At 24 weeks, the least-squares mean change in aLBM was -0·123 kg (95% CI -0·287 to 0·040) in the placebo group and 0·303 kg (0·135 to 0·470) in the LY group, a difference of 0·43 kg (95% CI 0·192 to 0·660; p<0·0001). Stair climbing time (four-step and 12-step tests), chair rise with arms, and fast gait speed improved significantly from baseline to week 24 with differences between LY and placebo of respectively -0·46 s (p=0·093), -1·28 s (p=0·011), -4·15 s (p=0·054), and 0·05 m/s (p=0·088). No effect was detected for other performance-based measures. Injection site reactions were recorded in nine (9%) patients given placebo and in 31 (30%) patients given LY (p<0·0001), and were generally mild, and led to treatment discontinuation in two patients given LY. INTERPRETATION: Our findings show LY treatment increases lean mass and might improve functional measures of muscle power. Although additional studies are needed to confirm these results, our data suggest LY should be tested for its potential ability to reduce the risk of falls or physical dependency in older weak fallers.