Background The evidence for non-invasive positive pressure ventilation (NIPPV) used in patients with severe stable chronic obstructive pulmonary disease (COPD) is insufficient. The aim of the meta-analysis was to assess the treatment effects of long-term NIPPV on gas change, lung function, health-related quality of life (HRQL), survival and mortality in severe stable COPD patients. Methods Randomized controlled trials (RCTs) and crossover studies comparing the treatment effects of NIPPV with conventional therapy were identified from electronic databases and reference lists from January 1995 to August 2010. Two reviewers independently assessed study quality. Data were combined using Review Manager 5.0. Both pooled effects and 95% confidence intervals were calculated. Results Five RCTs and one randomized crossover study with a total of 383 severe stable COPD patients were included. NIPPV improved gas change significantly when using a higher inspiratory positive airway pressures. The weighted mean difference (WMD) for the partial pressure of carbon dioxide in artery (PaCO2) was -3.52 (-5.26, -1.77) mmHg and for the partial pressure of oxygen in artery (PaO2) 2.84 (0.23, 5.44) mmHg. There were significant improvements in dyspnea and sleep quality, but gained no benefits on lung function. The standardized mean difference (SMD) for the forced expiratory volume in 1 second (FEV1) was 0.00 (0.29, 0.29). And the benefits for exercise tolerance, mood, survival and mortality remained unclear. Conclusions Patients with severe stable COPD can gain some substantial treatment benefits when using NIPPV, especially improvements in gas change, dyspnea and sleep quality. Studies of high methodological quality with large population, especially those based on a higher inspiratory positive airway pressures are required to provide more evidences. Chin Med J 2011;124(23):4063-4070