Northwest China is afflicted by frequent drought disasters, so accurate monitoring of drought conditions in the region is of great significance to agricultural production and economic development. This article uses meteorological station data from June 2001 to May 2020 and multi-source remote sensing data sets of precipitation, vegetation, soil and other environmental factors as data sources to establish a comprehensive drought monitoring model based on random forest algorithm. Then, the applicability of the model in the northwest China is investigated. The results show that the consistency rate of the drought levels in each month of the study period between the comprehensive drought index (CDI) output by the model and the meteorological comprehensive drought index (MCI) is above 75%. Moreover, the correlation coefficients between CDI and the standardized precipitation index (SPI-3), CDI and the standardized soil moisture index (SSI-1) of the meteorological stations are 0.318~0.726 and 0.173~0.433 respectively, both reaching extremely significant levels. In addition, CDI can accurately reflect the development of drought conditions in typical drought events in the northwest region. This study shows that the random forest method is applicable to the comprehensive drought monitoring in northwest China.