Summary Background Treatment with adjuvant trastuzumab for 1 year improves disease-free survival and overall survival in patients with human epidermal growth factor receptor 2 (HER2)-positive early breast cancer. We aimed to assess disease-free survival and overall survival after a median follow-up of 4 years for patients enrolled on the Herceptin Adjuvant (HERA) trial. Methods The HERA trial is an international, multicentre, randomised, open-label, phase 3 trial comparing treatment with trastuzumab for 1 and 2 years with observation after standard neoadjuvant, adjuvant chemotherapy, or both in patients with HER2-positive early breast cancer. The primary endpoint was disease-free survival. After a positive first interim analysis at a median follow-up of 1 year for the comparison of treatment with trastuzumab for 1 year with observation, event-free patients in the observation group were allowed to cross over to receive trastuzumab. We report trial outcomes for the 1-year trastuzumab and observation groups at a median follow-up of 48·4 months (IQR 42·0–56·5) and assess the effect of the extensive crossover to trastuzumab. Our analysis was by intention-to-treat. The HERA trial is registered with the European Clinical Trials Database, number 2005-002385-11. Findings The HERA trial population comprised 1698 patients randomly assigned to the observation group and 1703 to the 1-year trastuzumab group. Intention-to-treat analysis of disease-free survival showed a significant benefit in favour of patients in the 1-year trastuzumab group (4-year disease-free survival 78·6%) compared with the observation group (4-year disease-free survival 72·2%; hazard ratio HR 0·76; 95% CI 0·66–0·87; p<0·0001). Intention-to-treat analysis of overall survival showed no significant difference in the risk of death (4-year overall survival 89·3% vs 87·7%, respectively; HR 0·85; 95% CI 0·70–1·04; p=0·11). Overall, 885 patients (52%) of the 1698 patients in the observation group crossed over to receive trastuzumab, and began treatment at median 22·8 months (range 4·5–52·7) from randomisation. In a non-randomised comparison, patients in the selective-crossover cohort had fewer disease-free survival events than patients remaining in the observation group (adjusted HR 0·68; 95% CI 0·51–0·90; p=0·0077). Higher incidences of grade 3–4 and fatal adverse events were noted on 1-year trastuzumab than in the observation group. The most common grade 3 or 4 adverse events, each in less than 1% of patients, were congestive cardiac failure, hypertension, arthralgia, back pain, central-line infection, hot flush, headache, and diarrhoea. Interpretation Treatment with adjuvant trastuzumab for 1 year after chemotherapy is associated with significant clinical benefit at 4-year median follow-up. The substantial selective crossover of patients in the observation group to trastuzumab was associated with improved outcomes for this cohort. Funding F Hoffmann-La Roche, Michelangelo Foundation.