Background: The hospital environment is a reservoir for the transmission of microorganisms. The effect of improved cleaning on patient-centred outcomes remains unclear. We aimed to evaluate the effectiveness of an environmental cleaning bundle to reduce health care-associated infections in hospitals. Methods: The REACH study was a pragmatic, multicentre, randomised trial done in 11 acute care hospitals in Australia. Eligible hospitals had an intensive care unit, were classified by the National Health Performance Authority as a major hospital (public hospitals) or having more than 200 inpatient beds (private hospitals), and had a health-care-associated infection surveillance programme. The stepped-wedge design meant intervention periods varied from 20 weeks to 50 weeks. We introduced the REACH cleaning bundle, a multimodal intervention, focusing on optimising product use, technique, staff training, auditing with feedback, and communication, for routine cleaning. The primary outcomes were incidences of health-care-associated Staphylococcus aureus bacteraemia, Clostridium difficile infection, and vancomycin-resistant enterococci infection. The secondary outcome was the thoroughness of cleaning of frequent touch points, assessed by a fluorescent marking gel. This study is registered with the Australian and New Zealand Clinical Trial Registry, number ACTRN12615000325505. Findings: Between May 9, 2016, and July 30, 2017, we implemented the cleaning bundle in 11 hospitals. In the pre-intervention phase, there were 230 cases of vancomycin-resistant enterococci infection, 362 of S aureus bacteraemia, and 968 C difficile infections, for 3 534 439 occupied bed-days. During intervention, there were 50 cases of vancomycin-resistant enterococci infection, 109 of S aureus bacteraemia, and 278 C difficile infections, for 1 267 134 occupied bed-days. After the intervention, vancomycin-resistant enterococci infections reduced from 0·35 to 0·22 per 10 000 occupied bed-days (relative risk 0·63, 95% CI 0·41–0·97, p=0·0340). The incidences of S aureus bacteraemia (0·97 to 0·80 per 10 000 occupied bed-days; 0·82, 0·60–1·12, p=0·2180) and C difficile infections (2·34 to 2·52 per 10 000 occupied bed-days; 1·07, 0·88–1·30, p=0·4655) did not change significantly. The intervention increased the percentage of frequent touch points cleaned in bathrooms from 55% to 76% (odds ratio 2·07, 1·83–2·34, p<0·0001) and bedrooms from 64% to 86% (1·87, 1·68–2·09, p<0·0001). Interpretation: The REACH cleaning bundle was successful at improving cleaning thoroughness and showed great promise in reducing vancomycin-resistant enterococci infections. Our work will inform hospital cleaning policy and practice, highlighting the value of investment in both routine and discharge cleaning practice. Funding: National Health and Medical Research Council (Australia).