Background: Use of single-bed rooms for control of extended-spectrum β-lactamase (ESBL)-producing Enterobacteriaceae is under debate; the added value when applying contact precautions has not been shown. We aimed to assess whether an isolation strategy of contact precautions in a multiple-bed room was non-inferior to a strategy of contact precautions in a single-bed room for preventing transmission of ESBL-producing Enterobacteriaceae. Methods: We did a cluster-randomised, crossover, non-inferiority study on medical and surgical wards of 16 Dutch hospitals. During two consecutive study periods, either contact precautions in a single-bed room or contact precautions in a multiple-bed room were applied as the preferred isolation strategy for patients with ESBL-producing Enterobacteriaceae cultured from a routine clinical sample (index patients). Eligible index patients were aged 18 years or older, had no strict indication for barrier precautions in a single-bed room, had a culture result reported within 7 days of culture and before discharge, and had no wardmate known to be colonised or infected with an ESBL-producing Enterobacteriaceae isolate of the same bacterial species with a similar antibiogram. Hospitals were randomly assigned in a 1:1 ratio by computer to one of two sequences of isolation strategies, stratified by university or non-university hospital. Allocation was masked for laboratory technicians who assessed the outcomes but not for patients, treating doctors, and infection-control practitioners enrolling index patients. The primary outcome was transmission of ESBL-producing Enterobacteriaceae to wardmates, which was defined as rectal carriage of an ESBL-producing Enterobacteriaceae isolate that was clonally related to the index patient's isolate in at least one wardmate. The primary analysis was done in the per-protocol population, which included patients who were adherent to the assigned room type. A 10% non-inferiority margin for the risk difference was used to assess non-inferiority. This study is registered with Nederlands Trialregister, NTR2799. Findings: 16 hospitals were randomised, eight to each sequence of isolation strategies. All hospitals randomised to the sequence single-bed room then multiple-bed room and five of eight hospitals randomised to the sequence multiple-bed room then single-bed room completed both study periods and were analysed. From April 24, 2011, to Feb 27, 2014, 1652 index patients and 12 875 wardmates were assessed for eligibility. Of those, 693 index patients and 9527 wardmates were enrolled and 463 index patients and 7093 wardmates were included in the per-protocol population. Transmission of ESBL-producing Enterobacteriaceae to at least one wardmate was identified for 11 (4%) of 275 index patients during the single-bed room strategy period and for 14 (7%) of 188 index patients during the multiple-bed room strategy period (crude risk difference 3·4%, 90% CI −0·3 to 7·1). Interpretation: For patients with ESBL-producing Enterobacteriaceae cultured from a routine clinical sample, an isolation strategy of contact precautions in a multiple-bed room was non-inferior to a strategy of contact precautions in a single-bed room for preventing transmission of ESBL-producing Enterobacteriaceae. Non-inferiority of the multiple-bed room strategy might change the current single-bed room preference for isolation of patients with ESBL-producing Enterobacteriaceae and, thus, broaden infection-control options for ESBL-producing Enterobacteriaceae in daily clinical practice. Funding: Netherlands Organisation for Health Research and Development.