This study uses two forms of the Palmer Drought Severity Index (PDSI),namely the PDSI_TH (potential evapotranspiration estimated-by the Thornthwaite equation) and the PDSI_PM (potential evapotranspiration estimated by the FAO Penman-Monteith equation),to characterize the meteorological drought trends during 1960-2016 in the Loess Plateau (LP)and its four subregions.By designing a series of numerical experiments,we mainly investi-gated various climatic factors' contributions to the drought trends at annual,summer,and autumn time scales.Overall,the drying trend in the PDSI_TH is much larger than that in the PDSI_PM.The former is more sensitive to air temperature than precipitation,while the latter is the most sensitive to precipitation among all meteorological factors.Increasing temperature results in a decreasing trend (drying) in the PDSI_TH,which is further aggravated by de-creasing precipitation,jointly leading to a relatively severe drying trend.For the PDSI_PM that considers more comprehensive climatic factors,the drying trend is partly counteracted by the declining wind speed and solar radiation.Therefore,the PDSI_PM ultimately shows a much smaller drying trend in the past decades.