This study proposes an ensemble empirical mode decomposition (EEMD)-based multiscale entropy (EME) approach. The proposed model is used to analyze and gage variability of the annual precipitation series and spatially classify raingauges in Iran. For this end, historical annual precipitation data during 1960–2010 from 31 raingauges are decomposed using EEMD. Decomposed series of precipitation series present different periods and trends. Next, entropy concept is applied to the components obtained from EEMD to measure dispersion of multiscale components. It is observed that entropy values of intrinsic mode functions (IMFs) 1–5 and residual component show different behaviors. IMF 5 and residual components have highest values of entropy, whereas IMF 3 and 4 present highest entropy variation among all components. Based on spatial distribution of EME values, EME 3 and 1 have a downward variation from north to south, whereas EME 1 presents increasing variation. Spatial classification of raingauges is performed using EME values as input data to self-organizing map (SOM) and k-means clustering techniques. Finally, spatial structure of annual precipitation variation is investigated. It is observed that EME values have a downward trend with latitude, whereas it is observed that EME shows an upward relationship with longitude in Iran.