Abstract: Many studies have shown that drought severity is increasing over Yunnan-Guizhou (YG) region in southwestern China. And some studies stated that the deficit of precipitation was the main reason to recent droughts in southwestern China. Although precipitation is the main influence factor of drought formation, many studies have shown that evapotranspiration also plays a vital role in the progress of drought evolution. But how actual and potential evapotranspiration affect drought evolution is not well understood. In order to evaluate the temporal and spatial variations of the climate aridity in recent years as well as the roles of evapotranspiration in drought development in the YG region, monthly actual evapotranspiration and potential evapotranspiration data products during 2000-2014, including satellite-retrieved MODIS MOD16 data and GLDAS-Noah model simulation as well as the observed precipitation data were used to calculate 3 meteorological drought indices, i.e., standardized precipitation index (SPI), standardized reconnaissance drought index (RDIst) and evaporative stress index (ESI). With the Mann-Kendall trend test method, the 15-year temporal variations of SPI, RDIst and ESI in YG region are investigated. Besides, the severe drought that happened during 2009-2010 in YG region is further investigated as an example to analyze the role of actual evapotranspiration and potential evapotranspiration in drought evolution. According to the spatial distribution of annual average precipitation and the trend test results of 12-month SPI, there is a remarkable downtrend in middle Yunnan with an area of 12945 km2, indicating that areas with low annual precipitation are getting drier. The spatial patterns of trend tests for RDIst calculated based on different potential evapotranspiration are similar, consistent with that of SPI trend test result, both implying that there is a very significant drying trend in middle Yunnan and a slight drying in its surrounding areas, and the significant drying trend in the middle Yunnan is mainly controlled by the changes of precipitation. In terms of the trend of ESI, its spatial distributions based on MOD16 and GLDAS-Noah evapotranspiration data show great differences, but they both indicate an obvious drying trend in central Yunnan, which is similar to that of SPI and RDIst. Trend analyses of all 3 drought indices show that, the climate aridity in Guizhou Province is stable, while there is a considerable drying trend in middle Yunnan because of the joint effects of significant decline of precipitation and the remarkable increase of potential evapotranspiration. As the regional drying is not only controlled by the changes of precipitation but also affected by the changes of evapotranspiration, drying areal extents indicated by ESI and RDIst are larger than that indicated by SPI. During the 2009-2010 drought happened in YG region, actual evapotranspiration and potential evapotranspiration played little roles at the early stage, but with the evolution of drought, they greatly aggravated drought situations at the late stage. At the end of drought when precipitation returned to its normal condition, potential evapotranspiration kept positively biased for about 1-3 months, whereas actual evapotranspiration kept negatively biased for about 5 months, illustrating that actual evapotranspiration has a longer effect on drought evolution than potential evapotranspiration. The analysis of 2009-2010 severe drought also indicates the effectiveness of different drought indices, that is, ESI considers both actual and potential evapotranspiration, and is more reasonable than SPI and RDIst for monitoring drought evolution. So it can be concluded that when accurate actual evapotranspiration data are available, drought indices based on both actual evapotranspiration and potential evapotranspiration are preferable to those based on only potential evapotranspiration for drought monitoring. Unfortunately, there is a lack of reliable actual evapotranspiration products available most of time while there exist many widely used methods for potential evapotranspiration calculation, therefore when actual evapotranspiration products are not available, a practical option is using drought indices that take potential evapotranspiration into account.