BACKGROUND Every day, about 830 women die worldwide from preventable causes related to pregnancy and childbirth. Obstetric early warning scores have been proposed as a potential tool to reduce maternal morbidity and mortality, based on the identification of predetermined abnormal values in the vital signs or laboratory parameters, to generate a rapid and effective medical response. Several early warning scores have been developed for obstetrical patients, but the majority are the result of a clinical consensus rather than statistical analyses of clinical outcome measures (ie, maternal deaths). In 2013, the Intensive Care National Audit and Research Center Case Mix Program reported the first statistically validated early warning scoring system for pregnant women. OBJECTIVE We sought to assess the performance of the Intensive Care National Audit and Research Center Obstetric Early Warning Score in predicting death among pregnant women who required admission to the intensive care unit. STUDY DESIGN This retrospective cohort study included pregnant women admitted to the intensive care unit at a tertiary referral center from January 2006 through December 2011 in Colombia, a developing country, with direct and indirect obstetric-related conditions. The Obstetric Early Warning Score was calculated based on data collected during the first 24 hours of intensive care unit admission. The Obstetric Early Warning Score is calculated based on values of the following variables: systolic and diastolic blood pressure, respiratory rate, heart rate, fraction of inspired oxygen (FiO2) required to maintain an oxygen saturation ≥96%, temperature, and level of consciousness. The performance of the Obstetric Early Warning Score was evaluated using the area under the receiver operator characteristic curve. Outcomes selected were: maternal death, need for mechanical ventilation, and/or vasoactive support. Statistical methods included distribution appropriate univariate analyses and multivariate logistic regression. RESULTS During the study period, 50,897 births were recorded. There were 724 obstetric admissions to critical care, for an intensive care unit admission rate of 14.22 per 1000 deliveries. A total of 702 women were included in the study, with 29 (4.1%) maternal deaths, and a mortality ratio of 56.98 deaths per 100,000 live births. The most frequent causes of admission were direct, obstetric-related conditions (n = 534; 76.1%). The Obstetric Early Warning Score value was significantly higher in nonsurvivors than in survivors [12 (interquartile range 10-13) vs 7 (interquartile range 4-9); P < .001]. Peripartum women with normal values of Obstetric Early Warning Score had 0% mortality rate, while those with high Obstetric Early Warning Score values (>6) had a mortality rate of 6.3%. The area under the receiver operator characteristic curve of the Obstetric Early Warning Score in discrimination of maternal death was 0.84 (95% confidence interval, 0.75-0.92). The overall predictive value of the Obstetric Early Warning Score was better when the main cause of admission was directly related to pregnancy or the postpartum state. The area under the receiver operator characteristic curve of the score in conditions directly related to pregnancy and postpartum was 0.87 (95% confidence interval, 0.79-0.95), while in indirectly related conditions the area under the receiver operator characteristic curve was 0.77 (95% confidence interval, 0.58-0.96). CONCLUSION Although there are opportunities for improvement, Obstetric Early Warning Score obtained upon admission to the intensive care unit can predict survival in conditions directly related to pregnancy and postpartum. The use of early warning scores in obstetrics may be a highly useful approach in the early identification of women at an increased risk of dying.