Abstract： Objective To determine the prevalence of mortality and respiratory support therapy evolved for severe respiratory failure in pediatric intensive care units(PICU)in China. Method Two sessions of prospective, multicentric, and descriptively epidemiological survey of acute respiratory distress syndrome(ARDS)was carried out in year 2004 and 2006,and survey of alveolar hypoventilation respiratory failure(AHRF)in 2006 by 20 PICUs in China. Data about severity, outcome and respiratory therapy of critically ill patients, as well as PICU facilities were documented prospectively. PICU facilities, patients' mortality and respiratory therapy in different years were compared. Mortality risk factors were analyzed by multivariate logistic regression. Results Compared with those data in year 2004,the number of PICU bed and ventilator increased in year 2006.But,there was no increase in area per bed in year 2006.Proportion of critically ill patients in 2006 was significantly higher than that in 2004(69.0%vs.57.5%,x2=318,P<0.01).There were no significant difference in mortalities of critically ill patients between these two years. According to multivariate regression analysis, mortality was lowered in PICU with increase in heavy working load of physicians, evidenced by linear trend test P=0.013.Mortality of ARDS in year 2006 was significantly lower than that in 2004(39.7%vs 55.8%,x2=6.59,P<0.01).Whereas there was no significant difference in ventilation mode,tidal volume and airway pressure level for ventilated patients between two years.Conclusions With increasing the capacity of PICU there was a significant improvement of treatment as reflected by outcome of ARDS.Mortality of PICU patients was inversely correlated to the working load done by physicians.