BACKGROUND:Obesity paradox refers to lower mortality in subjects with higher body mass index (BMI), and has been documented under a variety of condition. However, whether obesity paradox exists in adults requiring mechanical ventilation in intensive critical units (ICU) remains controversial. METHODS:MEDLINE, EMBASE, China Biology Medicine disc (CBM) and CINAHL electronic databases were searched from the earliest available date to July 2017, using the following search terms: "body weight", "body mass index", "overweight" or "obesity" and "ventilator", "mechanically ventilated", "mechanical ventilation", without language restriction. Subjects were divided into the following categories based on BMI (kg/m2): underweight, 40 kg/m2. The primary outcome was mortality, and included ICU mortality, hospital mortality, short-term mortality (<6 months), and long-term mortality (6 months or beyond). Secondary outcomes included duration of mechanical ventilation, length of stay (LOS) in ICU and hospital. A random-effects model was used for data analyses. Risk of bias was assessed using the Newcastle-Ottawa quality assessment scale. RESULTS:A total of 15,729 articles were screened. The final analysis included 23 articles (199,421 subjects). In comparison to non-obese patients, obese patients had lower ICU mortality (odds ratio (OR) 0.88, 95% CI 0.0.84-0.92, I2 = 0%), hospital mortality (OR 0.83, 95% CI 0.74-0.93, I2 = 52%), short-term mortality (OR 0.81, 95% CI 0.74-0.88, I2 = 0%) as well as long-term mortality (OR 0.69, 95% CI 0.60-0.79, I2 = 0%). In comparison to subjects with normal BMI, obese patients had lower ICU mortality (OR 0.88, 95% CI 0.82-0.93, I2 = 5%). Hospital mortality was lower in severely obese and obese subjects (OR 0.71, 95% CI 0.53-0.94, I2 = 74%, and OR 0.80, 95% CI 0.73-0.89, I2 = 30%). Short-term mortality was lower in overweight and obese subjects (OR 0.82, 95% CI 0.75-0.90, I2 = 0%, and, OR 0.75, 95% CI 0.66-0.84, I2 = 8%, respectively). Long-term mortality was lower in severely obese, obese and overweight subjects (OR 0.39, 95% CI 0.18-0.83, and OR 0.63, 95% CI 0.46-0.86, I2 = 56%, and OR 0.66, 95% CI 0.57-0.77, I2 = 0%). All 4 mortality measures were higher in underweight subjects than in subjects with normal BMI. Obese subjects had significantly longer duration on mechanical ventilation than non-obese group (mean difference (MD) 0.48, 95% CI 0.16-0.80, I2 = 37%), In comparison to subjects with normal BMI, severely obese BMI had significantly longer time in mechanical ventilation (MD 1.10, 95% CI 0.38-1.83, I2 = 47%). Hospital LOS did not differ between obese and non-obese patients (MD 0.05, 95% CI -0.52 to 0.50, I2 = 80%). Obese patients had longer ICU LOS than non-obese patients (MD 0.38, 95% CI 0.17-0.59, I2 = 70%). Hospital LOS and ICU LOS did not differ significantly in subjects with different BMI status. CONCLUSIONS:In ICU patients receiving mechanical ventilation, higher BMI is associated with lower mortality and longer duration on mechanical ventilation.