The study is focused on the ecological-fragile and water-shortage region in Northwest China (NWC). A drought-heatwave cluster projection (DHCP) approach is developed based on Stepwise Cluster Analysis (SCA) and multi-level factor analysis (MFA). A multi-model ensemble consisting of 5 global climate models (GCMs) under two shared socioeconomic pathways (i.e., SSPs) is used in the approach. It could investigate the spatiotemporal characteristics of heatwaves, drought, and compound drought-heatwave events (CDHEs) in NWC. A precise projection is given by SCA in 48 stations and four indicators of CDHEs (i.e., CEN, CEDU, CEI, and CEM) are thus calculated. Finally, the individual and interactive effects of uncertainties (i.e., period, SSP, and GCM) are analyzed by MFA. Results show the SCA method can effectively reproduce precise projections for temperature and precipitation for NWC. The frequency, intensity, and duration of heatwaves increase significantly under SSP5-8.5. Drought will ease and then intensify projected by all SSPs. The long-term intensification trend of drought is more pronounced under SSP5-8.5. CDHEs will increase, especially under SSP5-8.5 by the 2090s. These parameters will increase to 4.59 events, 2.86 days per event, 16.7°C per event, and 4.69 days, respectively. Higher increases are found in southeastern Qinghai, northwestern Gansu, and northern Xinjiang. Period affects CDHEs projection mostly. The GCM selection and its interaction with period also affect the projection significantly. DHCP provides a comprehensive analysis of heatwaves and droughts. It is equally applicable to other water shortage areas and is expected to help better manage water resources.