Up-scaling method for inferring spatial information from a pixel scale to a basic unit scale has significant effects on aggregating results and decision-making. Therefore, developing appropriate methods to accurately up-scale spatial data is the key to infer useful drought information. The time series of vegetation temperature condition index (VTCI) drought monitoring results in Guanzhong Plain from early March to late May in the years from 2008 to 2013 were spatially transformed from a pixel scale to a basic unit scale by using the dominant class variability-weighted method (DCVW), arithmetic average variability-weighted method (AAVW) and median pixel variability-weighted method (MPVW) in the distributed mode and aggregated mode to obtain the aggregated VTCIs. The weighted VTCIs of winter wheat in main growth period were calculated, and the regression analysis between the weighted VTCIs and winter wheat yields was applied as references to evaluate up-scaling methods. The results showed that the regression analysis results of the three methods in the distributed up-scaling mode were generally better than those in the aggregated upscaling mode. The regression analysis results in the distributed up-scaling mode also indicated that the computing accuracy was high by DCVW and AAVW and was low by MPVW. DCVW in the distributed up-scaling mode was the most accurate method with the highest determination coefficient and the lowest estimated standard error, which were 0.64 and 289.97kg/hm2, respectively. The estimation yields of winter wheat which obtained by DCVW were very close to the levels of statistics yearbook of Shaanxi Province, indicating that the estimation precision of DCVW mehtod was high, and the method was robust. Overall, the method of DCVW in distributed up-scaling mode was the most reasonable approach to up-scale VTCIs in Guanzhong Plain from a pixel scale to a basic unit scale.