Primary Aim: Reduce audible cardiorespiratory (CR) alarm burden for all NICU patients by 20% between September 2015 and December 2017. Secondary Aim: Increase by 20% the proportion of Very Low Birth Weight (VLBW) infants spending 80% of their time within their target saturation range during the same epoch. Methods: We utilized the Model for Improvement as an underlying structure to address our aims. Our multidisciplinary improvement team included nursing leadership, physicians, frontline practitioners and hospital clinical engineers. Our team also included partners from the Institute for Medical Engineering at the Massachusetts Institute of Technology. We developed specific aims and measures and a key driver diagram (Figure 1). Multiple interventions were tested and implemented utilizing the Plan-Do-Study-Act (PDSA) model: (1) encouraged use of alarm silencing, January 2016; (2) staff education, November 2016; (3) implementation of oxygen saturation histograms, November 2017; (4) reinforcement of use of alarm pause feature through video education, February 2018; (5) change in high rate alarm limit from 200 to 210 for infants above 34 weeks, March 2018. In addition, in December 2017, we upgraded our CR monitor software to enable near-real time streaming of waveform and alarm data from all NICU beds to a server at MIT. Results: We measured a mean of 186,000 alarms per month in the NICU with wide month-to-month variation (SD 27,900). The effect of our interventions on the overall CR alarm rate is illustrated in Figure 2. Our baseline rate of 166 CR alarms per patient day was reduced to 132. We noted modest improvements after our educational efforts (PDSA #1 and #2). Analysis of the unusually high alarm rate in September 2016 revealed it to be driven by two individual patients. We utilized a retrospective cohort to develop algorithms for processing saturation data to determine time spent in target range and implemented PDSA #3. We performed analysis of our advisory (yellow) tachycardia alarms and determined that 80% resolved before reaching a critical threshold. This prompted implementation of PDSA#4, which reduced our average burden by nine alarms per patient per day (Figure 3). Discussion: The measured CR alarm burden in this single center quality improvement study is immense. As we began to explore alarm rates in the NICU, it became clear that individual patient factors have an extraordinary impact. While our educational packages and process changes may have had a modest effect on our burden, we did not observe sustained improvement. Moving forward we will focus on identification of individual patient and work flow factors that affect alarm rates and identify hard wired changes that will affect care processes. With our established cohesive team and system of rigorous data tracking we can perform deeper data analysis and position our team for success.