Serratia marcescens acquisition is a common problem in intensive care units (ICUs). Following an initial outbreak in 2017 with ongoing endemicity, this study aimed to analyze the impact of behavioral interventions and sink removals on S. marcescens incidence in a tertiary-care ICU. We conducted a quasi-experimental, interventional study including patients with a positive screening or clinical culture for S. marcescens , from 48 h (D2) after ICU admission to 14 days after ICU discharge. A sub-analysis considered patients positive for S. marcescens from ICU admission (D0) to 14 days after ICU discharge. Multivariate Poisson regression analyses were performed. Between January 2014 and December 2022, 167 cases of S. marcescens infection or colonization were identified (respiratory samples, 71%). Despite the presence of an aquatic reservoir, we found that neither behavioral nor architectural interventions (sink removal) reduced significantly S. marcescens incidence, yielding incidence ratios of 1.02 [95%CI 0.33-3.11] and 4.25 [95%CI 0.59-30.56], respectively. However, an association was observed with administration of selective oral decontamination (SOD) in the sub-cohort (OR 1.01; 95%CI 1.00-1.03). Behavioral change interventions and transition to a waterless ICU did not control endemic, polyclonal S. marcescens occurrence. The selective pressure exercised by SOD may have reduced the effectiveness of waterless care.