To explore the diagnostic value of mechanical power (MP) in patients with moderate to severe acute respiratory distress syndrome (ARDS) based on the Medical Information Mart for Intensive Care-IIIv1.4 (MIMIC-III v1.4). The information of ARDS patients undergoing invasive mechanical ventilation for no less than 48 hours who were hospitalized at Beth Israel Deaconess Medical Center in Boston, Massachusetts from June 2001 to October 2012 in the MIMIC-III v1.4 were collected. The demographics of patients, disease severity scores, ARDS etiology, prognostic indicators, pre-ventilation arterial blood gas analysis and respiratory parameters within 48 hours of ventilation were extracted. According to the lowest oxygenation index (PaO2/FiO2) before ventilation, the patients were divided into mild to moderate ARDS group (> 150 mmHg, 1 mmHg≈0.133 kPa) and moderate to severe ARDS group (≤ 150 mmHg), and the differences in baseline characteristics between the two groups were compared. The independent predictors associated with the severity of ARDS were analyzed using Logistic regression. The receiver operator characteristic curve (ROC curve) was plotted. The area under ROC curve (AUC) was calculated to evaluate the diagnostic value of MP for moderate to severe ARDS. The Youden index was used to determine the diagnostic threshold of MP for moderate to severe ARDS. According to the cut-off value of MP based on Youden index, all ARDS patients were divided into high and low MP groups. Kaplan-Meier survival curve was used to analyze the 28-day survival status of patients. A total of 403 ARDS patients were enrolled in the study, including 107 subjects with mild to moderate ARDS and 296 with moderate to severe ARDS. There were significant differences in age, sequential organ failure assessment (SOFA) score, the lowest PaO2/FiO2 before ventilation, the last PaO2/FiO2 before ventilation, 28-day mortality, the length of intensive care unit (ICU) stay, duration of mechanical ventilation, lung dynamic compliance (Cdyn) in the second 24 hours of ventilation and positive end-expiratory pressure (PEEP), plateau pressure (Pplat), driving pressure (ΔP), respiratory rate (RR), lung static compliance (Cst), MP, inspired fraction of oxygen (FiO2) within 48 hours of ventilation between the two groups. After adjusting variables such as age, SOFA score, the last PaO2/FiO2 before ventilation, and related respiratory mechanics parameters, multivariate Logistic regression analysis showed that higher ΔP, PEEP and MP, and lower last PaO2/FiO2 before ventilation were independently associated with moderate to severe ARDS [odds ratio (OR) and 95% confidence interval (95%CI) was 1.137 (1.032-1.252), 1.333 (1.139-1.561), 1.102 (1.030-1.179), and 0.996 (0.993-0.998), respectively, all P < 0.01]. The ROC curve analysis showed that the best cut-off value of MP for the diagnosis of moderate to severe ARDS was 18.1 J/min with sensitivity of 81.42% and specificity of 60.75%, and the AUC was 0.745 (95%CI was 0.690-0.799). According to the cut-off value of MP obtained by ROC curve, all ARDS patients were divided into high MP group (> 18.1 J/min) and low MP group (≤ 18.1 J/min). The Kaplan-Meier survival curve showed that the 28-day cumulative survival rate in the high MP group was significantly lower than that in the low MP group (73.8% vs. 85.1%; Log-Rank test: χ2 = 5.660, P = 0.017). MP is an independent predictor of the severity of ARDS, and it can be used to diagnose moderate to severe ARDS.