BACKGROUND: Healthcare-associated infections (HAIs), or nosocomial infections, represent a significant burden in terms of mortality, morbidity, length of stay and costs for patients hospitalized in intensive care units (ICUs). Surveillance systems are recommended by national and international institutions to gather data on HAIs in order to develop and evaluate interventions that reduce the risk of HAIs. STUDY DESIGN: Here we describe the methodology and the results of the surveillance system implemented in the ICU of the Policlinico Umberto I, a large teaching hospital in Rome, from April 2016 to October 2018. METHODS: The multimodal infection surveillance system integrates four different approaches: i) active surveillance of inpatients; ii) environmental microbiological surveillance; iii) surveillance of isolated microorganisms; and iv) behavioral surveillance of healthcare personnel. Data were collected on catheter-related bloodstream infections, ventilation-associated pneumonia, catheter-associated urinary tract infections and primary bloodstream infections that developed in patients after 48 h in the ICU. For environmental surveillance 14 points were selected for sampling (i.e. bed edges, medication carts, PC keyboards, sink faucets). The system of active surveillance of HAIs also included surveillance of microorganisms, consisting of the molecular genotyping of bacterial isolates by pulsed-field gel electrophoresis (PFGE). From 1 November 2016, monitoring of compliance with guidelines for hand hygiene (HH) and proper glove or gown use by healthcare personnel was included in the surveillance system. After the first six months (baseline phase), a multimodal intervention to improve adherence to guidelines by healthcare personnel was conducted with the ICU staff. RESULTS: Overall, 773 patients were included in the active surveillance. The overall incidence rate of device-related HAIs was 14.1 (95% CI: 12.2-16.3) per 1000 patient-days. The monthly device-related HAI incident rate showed a decreasing trend over time, with peaks of incidence becoming progressively lower. The most common bacterial isolates were Klebsiella pneumoniae (20.7%), Acinetobacter baumannii (17.2%), Pseudomonas aeruginosa (13.4%) and Staphylococcus aureus (5.4%). Acinetobacter baumannii and Klebsiella pneumoniae showed the highest proportion of isolates with a multidrug-resistant profile. A total of 819 environmental samples were collected, from which 305 bacterial isolates were retrieved. The most frequent bacterial isolates were Acinetobacter baumannii (27.2%), Staphylococcus aureus (12.1%), Enterococcus faecalis (11.1%), Klebsiella pneumoniae (5.2%) and Pseudomonas aeruginosa (4.7%). All Acinetobacter baumannii, Pseudomonas aeruginosa and Klebsiella pneumoniae environmental isolates were at least multidrug-resistant. Genotyping showed a limited number of major PFGE patterns for both clinical and environmental isolates of Klebsiella pneumoniae and Acinetobacter baumannii. Behavioral compliance rates significantly improved from baseline to post-intervention phase. CONCLUSIONS: By integrating information gathered from active surveillance, environmental microbiological surveillance, surveillance of bacterial isolates and behavioral surveillance of healthcare personnel, the multimodal infection surveillance system returned a precise and detailed view of the infectious risk and microbial ecology of the ICU.