To observe the treatment strategy and its changes in mechanical ventilation (MV) in a single medical center. Five hundred and two patients undergoing MV for at least 24 hours from January 1994 to December 1997 (control group) and from January 2004 to December 2006 (study group) in a total of 1 090 patients who were admitted to intensive care unit (ICU) Fuxing Hospital, Capital Medical University during the 2 periods were investigated. Datas including causes for the initiation of MV, ventilator modes and treatment parameters, weaning methods, and prognosis of patients were collected. The total incidence of MV was 46.1% (502/1 090). The incidence of MV in control group was 48.9% (184/376), and that in study group was 44.5% (318/714), respectively. The main causes for MV of 502 patients were pneumonia 18.3% (92 cases), acute exacerbation of chronic obstructive pulmonary disease (AECOPD) 16.3% (82 cases), postoperation 13.7% (69 cases), coma 14.1% (71 cases) , and acute respiratory distress syndrome (ARDS) 12.7% (64 cases). The initial ventilator mode: 59.8% (110/184) or 23.0%(73/318) in control or study group was assist/control ventilation (A/C), and 57.2% (182/318) or 20.7%(38/184) in study or control group was pressure support ventilation (PSV), and there was significant difference between the two groups (both P<0.01). The use of noninvasive ventilation (NPPV) in study group was obviously increased compared with control group [10.4% (33/318) vs. 3.8% (7/184), P<0.01]. The mean pressure level of pressure support (PS) of all patients was 14.0 cm H(2)O (1 cm H(2)O=0.098 kPa), the mean positive end-expiratory pressure (PEEP) of both groups was 5.0 cm H(2)O. Compared with control group, PEEP (cm H(2)O) level in patients with ARDS was significantly higher (8.0 vs. 6.0, P<0.01) and volume tidal (V(T), ml) was significantly lower (400 vs. 550, P<0.01) in study group. The most frequently used weaning methods of both groups were T-piece, T-piece+PSV and PSV. The use of T-piece in study group was significantly higher than that in control group [84.4% (184/218) vs. 35.1% (40/114), P<0.01], and PSV was lower than that in control group [2.8% (6/218) vs. 29.8% (34/114), P<0.01]. The total mortality of MV patients in two groups in ICU was 49.6%(249/502). There was no significant difference of the mortality between study group and control group (54.6% vs. 55.4%, P=0.887). The ventilator modes and settings had been changed in a single medical center in the past 10 years. It is speculated that the changes are related with the results observed in some multicenter randomized controlled trials (RCTs).