Under the framework of an project ‘Building Asian Climate Change Scenarios by Multi-Regional Climate Models Ensemble’, the ability of eight regional climate models and two fine-resolution global climate models to reproduce late 20th century (1981–2000) precipitation climatology is assessed. Future precipitation change (2041–2060) under the A1B scenario is also quantified by applying four different ensemble methods: equal weighting, weighted mean (WM), reliability ensemble averaging (REA) and performance-based ensemble averaging, after applying fourfold cross-validation using observation and multi-model-simulated precipitation. The results indicate that the ensemble of simulated precipitation outperforms any single RCM in many aspects. Among the four ensemble approaches, the WM and REA methods show better skill in improving the simulation results, and are used for ensemble prediction of regional climate in Asia. Under the A1B scenario, the WM method estimates future precipitation change of approximately 0.2 mm day−1 with less precipitation in northern and western China and northern India, and more precipitation in most other areas in Asia. The future annual precipitation will decrease by 0.1-0.5 mm day−1 in northern India, Pakistan and the central area of southern China. No significant change is found over eastern Kazakhstan, Mongolia, north-central and western China.