Background Cabozantinib is an oral inhibitor of tyrosine kinases including MET, VEGFR, and AXL. The randomised phase 3 METEOR trial compared the efficacy and safety of cabozantinib versus the mTOR inhibitor everolimus in patients with advanced renal cell carcinoma who progressed after previous VEGFR tyrosine-kinase inhibitor treatment. Here, we report the final overall survival results from this study based on an unplanned second interim analysis. Methods In this open-label, randomised phase 3 trial, we randomly assigned (1:1) patients aged 18 years and older with advanced or metastatic clear-cell renal cell carcinoma, measurable disease, and previous treatment with one or more VEGFR tyrosine-kinase inhibitors to receive 60 mg cabozantinib once a day or 10 mg everolimus once a day. Randomisation was done with an interactive voice and web response system. Stratification factors were Memorial Sloan Kettering Cancer Center risk group and the number of previous treatments with VEGFR tyrosine-kinase inhibitors. The primary endpoint was progression-free survival as assessed by an independent radiology review committee in the first 375 randomly assigned patients and has been previously reported. Secondary endpoints were overall survival and objective response in all randomly assigned patients assessed by intention-to-treat. Safety was assessed per protocol in all patients who received at least one dose of study drug. The study is closed for enrolment but treatment and follow-up of patients is ongoing for long-term safety evaluation. This trial is registered with ClinicalTrials.gov, number NCT01865747. Findings Between Aug 8, 2013, and Nov 24, 2014, 658 patients were randomly assigned to receive cabozantinib (n=330) or everolimus (n=328). The median duration of follow-up for overall survival and safety was 18·7 months (IQR 16·1–21·1) in the cabozantinib group and 18·8 months (16·0–21·2) in the everolimus group. Median overall survival was 21·4 months (95% CI 18·7–not estimable) with cabozantinib and 16·5 months (14·7–18·8) with everolimus (hazard ratio [HR] 0·66 [95% CI 0·53–0·83]; p=0·00026). Cabozantinib treatment also resulted in improved progression-free survival (HR 0·51 [95% CI 0·41–0·62]; p<0·0001) and objective response (17% [13–22] with cabozantinib vs 3% [2–6] with everolimus; p<0·0001) per independent radiology review among all randomised patients. The most common grade 3 or 4 adverse events were hypertension (49 [15%] in the cabozantinib group vs 12 [4%] in the everolimus group), diarrhoea (43 [13%] vs 7 [2%]), fatigue (36 [11%] vs 24 [7%]), palmar-plantar erythrodysaesthesia syndrome (27 [8%] vs 3 [1%]), anaemia (19 [6%] vs 53 [17%]), hyperglycaemia (3 [1%] vs 16 [5%]), and hypomagnesaemia (16 [5%] vs none). Serious adverse events grade 3 or worse occurred in 130 (39%) patients in the cabozantinib group and in 129 (40%) in the everolimus group. One treatment-related death occurred in the cabozantinib group (death; not otherwise specified) and two occurred in the everolimus group (one aspergillus infection and one pneumonia aspiration). Interpretation Treatment with cabozantinib increased overall survival, delayed disease progression, and improved the objective response compared with everolimus. Based on these results, cabozantinib should be considered as a new standard-of-care treatment option for previously treated patients with advanced renal cell carcinoma. Patients should be monitored for adverse events that might require dose modifications. Funding Exelixis Inc.