This study mostly aimed to modify the calculation process of the standardized precipitation index (SPI) to survey the droughts in arid and semi-arid areas. The selection of the best probability distribution functions, selection of the most appropriate probabilistic estimates to assign SPI values to zero precipitations, and recommendation of an appropriate precipitation time series arrangement at different time scales were focused attention to modify the SPI calculation process. The results demonstrated that the probability distribution of the generalized extreme values at monthly, seasonal and annual time scales was the best alternative to calculate the SPI than the default gamma distribution. To assign SPI values to zero precipitation at precipitation time series of arid and semi-arid areas which have a specific seasonal precipitation regime, the statistical center of mass of zero distribution method yielded better results than the maximum likelihood estimation method. Finally, to arrange precipitation time series of arid and semi-arid areas, it was proposed to consider the individual months and seasons as an independent statistical population rather than to consider the entire monthly rainfall time series of these areas as an independent statistical population. This is because if this type of rainfall time series arrangement is not selected, the dry months are shown to be less important while the wet months more important when calculating the SPI.