Background The emergence of highly active novel agents has led some to question the role of autologous haematopoietic stem-cell transplantation (HSCT) and subsequent consolidation therapy in newly diagnosed multiple myeloma. We therefore compared autologous HSCT with bortezomib–melphalan–prednisone (VMP) as intensification therapy, and bortezomib–lenalidomide–dexamethasone (VRD) consolidation therapy with no consolidation. Methods In this randomised, open-label, phase 3 study we recruited previously untreated patients with multiple myeloma at 172 academic and community practice centres of the European Myeloma Network. Eligible patients were aged 18–65 years, had symptomatic multiple myeloma stage 1–3 according to the International Staging System (ISS), measurable disease (serum M protein >10 g/L or urine M protein >200 mg in 24 h or abnormal free light chain [FLC] ratio with involved FLC >100 mg/L, or proven plasmacytoma by biopsy), and WHO performance status grade 0–2 (grade 3 was allowed if secondary to myeloma). Patients were first randomly assigned (1:1) to receive either four 42-day cycles of bortezomib (1·3 mg/m2 administered intravenously or subcutaneously on days 1, 4, 8, 11, 22, 25, 29, and 32) combined with melphalan (9 mg/m2 administered orally on days 1–4) and prednisone (60 mg/m2 administered orally on days 1–4) or autologous HSCT after high-dose melphalan (200 mg/m2), stratified by site and ISS disease stage. In centres with a double HSCT policy, the first randomisation (1:1:1) was to VMP or single or double HSCT. Afterwards, a second randomisation assigned patients to receive two 28-day cycles of consolidation therapy with bortezomib (1·3 mg/m2 either intravenously or subcutaneously on days 1, 4, 8, and 11), lenalidomide (25 mg orally on days 1–21), and dexamethasone (20 mg orally on days 1, 2, 4, 5, 8, 9, 11, and 12) or no consolidation; both groups received lenalidomide maintenance therapy (10 mg orally on days 1–21 of a 28-day cycle). The primary outcomes were progression-free survival from the first and second randomisations, analysed in the intention-to-treat population, which included all patients who underwent each randomisation. All patients who received at least one dose of study drugs were included in the safety analyses. This study is registered with the EU Clinical Trials Register (EudraCT 2009-017903-28) and ClinicalTrials.gov (NCT01208766), and has completed recruitment. Findings Between Feb 25, 2011, and April 3, 2014, 1503 patients were enrolled. 1197 patients were eligible for the first randomisation, of whom 702 were assigned to autologous HSCT and 495 to VMP; 877 patients who were eligible for the first randomisation underwent the second randomisation to VRD consolidation (n=449) or no consolidation (n=428). The data cutoff date for the current analysis was Nov 26, 2018. At a median follow-up of 60·3 months (IQR 52·2–67·6), median progression-free survival was significantly improved with autologous HSCT compared with VMP (56·7 months [95% CI 49·3–64·5] vs 41·9 months [37·5–46·9]; hazard ratio [HR] 0·73, 0·62–0·85; p=0·0001). For the second randomisation, the number of events of progression or death at data cutoff was lower than that preplanned for the final analysis; therefore, the results from the second protocol-specified interim analysis, when 66% of events were reached, are reported (data cutoff Jan 18, 2018). At a median follow-up of 42·1 months (IQR 32·3–49·2), consolidation therapy with VRD significantly improved median progression-free survival compared with no consolidation (58·9 months [54·0–not estimable] vs 45·5 months [39·5–58·4]; HR 0·77, 0·63–0·95; p=0·014). The most common grade ≥3 adverse events in the autologous HSCT group compared to the VMP group included neutropenia (513 [79%] of 652 patients vs 137 [29%] of 472 patients), thrombocytopenia (541 [83%] vs 74 [16%]), gastrointestinal disorders (80 [12%] vs 25 [5%]), and infections (192 [30%] vs 18 [4%]). 239 (34%) of 702 patients in the autologous HSCT group and 135 (27%) of 495 in the VMP group had at least one serious adverse event. Infection was the most common serious adverse event in each of the treatment groups (206 [56%] of 368 and 70 [37%] of 189). 38 (12%) of 311 deaths from first randomisation were likely to be treatment related: 26 (68%) in the autologous HSCT group and 12 (32%) in the VMP group, most frequently due to infections (eight [21%]), cardiac events (six [16%]), and second primary malignancies (20 [53%]). Interpretation This study supports the use of autologous HSCT as intensification therapy and the use of consolidation therapy in patients with newly diagnosed multiple myeloma, even in the era of novel agents. The role of high-dose chemotherapy needs to be reassessed in future studies, in particular in patients with undetectable minimal residual disease after four-drug induction regimens including a monoclonal antiboby combined with an immunomodulatory agent and a proteasome inhibitor plus dexamethasone. Funding Janssen and Celgene.