Although total storage deficit index (TSDI) is a well-known GRACE-based drought index, it is not efficient for the basins with high consumption because water harvesting in these areas affect GRACE signal. To overcome this limitation, the modified total storage deficit index (MTSDI) is introduced in this paper. To develop MTSDI, residuals of the signal were used instead of total signal. The proposed approach was applied to monitor drought in the Markazi Basin, Iran, during 2002–2016. Based on the obtained results, TSDI detected a moderate drought event in 2012 and a long-term severe drought from 2013 to 2016, which is not compatible with the recorded results of standardized precipitation index (SPI), Standardized Precipitation Evapotranspiration Index (SPEI). Meanwhile MTSDI identified two drought events in 2008–09 and 2010–11 that coincided with droughts detected using SPI and SPEI. Results revealed that MTSDI is highly correlated with SPI and SPEI at a 12 month-scale, showing a correlation coefficient of 0.75 and 0.62, respectively. On the other hand, the coefficient of correlation between TSDI and SPI12 was 0.42 and between TSDI and SPEI12 was 0.26. A pixel-by-pixel analysis showed MTSDI and SPI/SPEI had a significant correlation in most areas of the basin. Further, the investigation of droughts in Markazi Basin revealed that dry years coincide with the occurrence of strong La Niña events. Overall, the results indicated a good potential of GRACE observations for developing a drought monitoring system, which its performance can be enhanced with respect to large-scale climate signals.