© 2019 Elsevier Ltd Background: Odanacatib, a cathepsin K inhibitor, reduces bone resorption while maintaining bone formation. Previous work has shown that odanacatib increases bone mineral density in postmenopausal women with low bone mass. We aimed to investigate the efficacy and safety of odanacatib to reduce fracture risk in postmenopausal women with osteoporosis. Methods: The Long-term Odanacatib Fracture Trial (LOFT) was a multicentre, randomised, double-blind, placebo-controlled, event-driven study at 388 outpatient clinics in 40 countries. Eligible participants were women aged at least 65 years who were postmenopausal for 5 years or more, with a femoral neck or total hip bone mineral density T-score between −2·5 and −4·0 if no previous radiographic vertebral fracture, or between −1·5 and −4·0 with a previous vertebral fracture. Women with a previous hip fracture, more than one vertebral fracture, or a T-score of less than −4·0 at the total hip or femoral neck were not eligible unless they were unable or unwilling to use approved osteoporosis treatment. Participants were randomly assigned (1:1) to either oral odanacatib (50 mg once per week) or matching placebo. Randomisation was done using an interactive voice recognition system after stratification for previous radiographic vertebral fracture, and treatment was masked to study participants, investigators and their staff, and sponsor personnel. If the study completed before 5 years of double-blind treatment, consenting participants could enrol in a double-blind extension study (LOFT Extension), continuing their original treatment assignment for up to 5 years from randomisation. Primary endpoints were incidence of vertebral fractures as assessed using radiographs collected at baseline, 6 and 12 months, yearly, and at final study visit in participants for whom evaluable radiograph images were available at baseline and at least one other timepoint, and hip and non-vertebral fractures adjudicated as being a result of osteoporosis as assessed by clinical history and radiograph. Safety was assessed in participants who received at least one dose of study drug. The adjudicated cardiovascular safety endpoints were a composite of cardiovascular death, myocardial infarction, or stroke, and new-onset atrial fibrillation or flutter. Individual cardiovascular endpoints and death were also assessed. LOFT and LOFT Extension are registered with ClinicalTrials.gov (number NCT00529373) and the European Clinical Trials Database (EudraCT number 2007-002693-66). Findings: Between Sept 14, 2007, and Nov 17, 2009, we randomly assigned 16 071 evaluable patients to treatment: 8043 to odanacatib and 8028 to placebo. After a median follow-up of 36·5 months (IQR 34·43–40·15) 4297 women assigned to odanacatib and 3960 assigned to placebo enrolled in LOFT Extension (total median follow-up 47·6 months, IQR 35·45–60·06). In LOFT, cumulative incidence of primary outcomes for odanacatib versus placebo were: radiographic vertebral fractures 3·7% (251/6770) versus 7·8% (542/6910), hazard ratio (HR) 0·46, 95% CI 0·40–0·53; hip fractures 0·8% (65/8043) versus 1·6% (125/8028), 0·53, 0·39–0·71; non-vertebral fractures 5·1% (412/8043) versus 6·7% (541/8028), 0·77, 0·68–0·87; all p<0·0001. Combined results from LOFT plus LOFT Extension for cumulative incidence of primary outcomes for odanacatib versus placebo were: radiographic vertebral fractures 4·9% (341/6909) versus 9·6% (675/7011), HR 0·48, 95% CI 0·42–0·55; hip fractures 1·1% (86/8043) versus 2·0% (162/8028), 0·52, 0·40–0·67; non-vertebral fractures 6·4% (512/8043) versus 8·4% (675/8028), 0·74, 0·66–0·83; all p<0·0001. In LOFT, the composite cardiovascular endpoint of cardiovascular death, myocardial infarction, or stroke occurred in 273 (3·4%) of 8043 patients in the odanacatib group versus 245 (3·1%) of 8028 in the placebo group (HR 1·12, 95% CI 0·95–1·34; p=0·18). New-onset atrial fibrillation or flutter occurred in 112 (1·4%) of 8043 patients in the odanacatib group versus 96 (1·2%) of 8028 in the placebo group (HR 1·18, 0·90–1·55; p=0·24). Odanacatib was associated with an increased risk of stroke (1·7% [136/8043] vs 1·3% [104/8028], HR 1·32, 1·02–1·70; p=0·034), but not myocardial infarction (0·7% [60/8043] vs 0·9% [74/8028], HR 0·82, 0·58–1·15; p=0·26). The HR for all-cause mortality was 1·13 (5·0% [401/8043] vs 4·4% [356/8028], 0·98–1·30; p=0·10). When data from LOFT Extension were included, the composite of cardiovascular death, myocardial infarction, or stroke occurred in significantly more patients in the odanacatib group than in the placebo group (401 [5·0%] of 8043 vs 343 [4·3%] of 8028, HR 1·17, 1·02–1·36; p=0·029, as did stroke (2·3% [187/8043] vs 1·7% [137/8028], HR 1·37, 1·10–1·71; p=0·0051). Interpretation: Odanacatib reduced the risk of fracture, but was associated with an increased risk of cardiovascular events, specifically stroke, in postmenopausal women with osteoporosis. Based on the overall balance between benefit and risk, the study's sponsor decided that they would no longer pursue development of odanacatib for treatment of osteoporosis. Funding: Merck Sharp & Dohme Corp, a subsidiary of Merck & Co, Inc, Kenilworth, NJ, USA.