BACKGROUND: Previously, we assessed selective digestive tract decontamination (SDD) and selective oropharyngeal decontamination (SOD) on survival and prevention of bacteraemia in patients in intensive-care units. In this analysis, we aimed to assess effectiveness of these interventions for prevention of respiratory tract colonisation and bacteraemia with highly resistant microorganisms acquired in intensive-care units. METHODS: We did an open-label, clustered group-randomised, crossover study in 13 intensive-care units in the Netherlands between May, 2004, and July, 2006. Participants admitted to intensive-care units with an expected duration of mechanical ventilation of more than 48 h or an expected stay of more than 72 h received SOD (topical tobramycin, colistin, and amphotericin B in the oropharynx), SDD (SOD antibiotics in the oropharynx and stomach plus 4 days' intravenous cefotaxime), or standard care. The computer-randomised order of study regimens was applied by an independent clinical pharmacist who was masked to intensive-care-unit identity. We calculated crude odds ratios (95% CI) for rates of bacteraemia or respiratory tract colonisation with highly resistant microorganisms in patients who stayed in intensive-care units for more than 3 days (ie, acquired infection). This trial is registered at http://isrctn.org, number ISRCTN35176830. FINDINGS: Data were available for 5927 (>99%) of 5939 patients, of whom 5463 (92%) were in intensive-care units for more than 3 days. 239 (13%) of 1837 patients in standard care acquired bacteraemia after 3 days, compared with 158 (9%) of 1758 in SOD (odds ratio 0.66, 95% CI 0.53-0.82), and 124 (7%) of 1868 in SDD (0.48, 0.38-0.60). Eight patients acquired bacteraemia with highly resistant microorganisms during SDD, compared with 18 patients (with 19 episodes) during standard care (0.41, 0.18-0.94; rate reduction [RR] 59%, absolute risk reduction [ARR] 0.6%) and 20 during SOD (0.37, 0.16-0.85; RR 63%, ARR 0.7%). Of the patients staying in intensive-care units for more than 3 days, we obtained endotracheal aspirate cultures for 881 (49%) patients receiving standard care, 886 (50%) receiving SOD, and 828 (44%) receiving SDD. 128 (15%) patients acquired respiratory tract colonisation with highly resistant microorganisms during standard care, compared with 74 (8%) during SDD (0.58, 0.43-0.78; RR 38%, ARR 5.5%) and 88 (10%) during SOD (0.65, 0.49-0.87; RR 32%, ARR 4.6%). Acquired respiratory tract colonisation with Gram-negative bacteria or cefotaxime-resistant and colistin-resistant pathogens was lowest during SDD. INTERPRETATION: Widespread use of SDD and SOD in intensive-care units with low levels of antibiotic resistance is justified. FUNDING: None.