Coconut is one of the major plantation crops in Sri Lanka which is more sensitive to the climatic change. During past few years in Sri Lanka the climate change has been extensively discussed, because it is important to investigate the climate extremes in order to best address their impacts in the agriculture section in future. The main objective in this study is to identify rainfall patterns and forecasting the extreme drought events in major coconut growing agro-ecological regions (AERs) during the monsoon seasons and Yala/Maha seasons. Daily rainfall data from 1932 to 2011in 14 rainfall stations from six AERs, DL3, WL2, WL2b, WL3, IL1 and IL3 were acquired from the climate database from Biometry division in Coconut Research Institute. Climate indices provide valuable information contained in daily rainfall data. In this study five descriptive extreme precipitation indices, which are defined by Expert Team on Climatic Change Detection and Indices (ETCCDI) were used to identify characteristic trend of extreme events. There are two types of extreme rainfall indices interested in this study, those are extreme frequency (R10mm, R20mm, R95p, R99p) and extreme intensity (Rx1day). Mann Kendall trend test used to identify trend of each indices. It was noted that trend analysis of extreme precipitation indices were revealed that the occurrence of drought events had a significant increase in WL2b region, Horakelle, Palugaswewa stations in IL1 region and Ambalantota station in the DL3 region. Standard Precipitation Index (SPI) was used as the drought-monitoring tool with three different time scales, those are SPI3, SPI6, and SPI12 time scale. SPI12 time scale was used to obtain the historical hydrological drought event. It was noted that all stations in WL2b region had highest drought duration and drought severity while Kekanadura station had highest drought duration with-96.58 drought severity with 35 month. SPI3 and SPI6 time scale values were used to analysis the drought events in monsoon seasons and Yala/Maha seasons respectively. To forecast future drought events in each stations Autoregressive Integrated Moving Average (ARIMA) and Seasonal Auto Regressive Integrated Moving Average (SARIMA) models were fitted in monsoon seasons and Yala/Maha season. To identify drought areas with similar characteristics in different seasons, cluster analysis for SPI3 and SPI6 time scale were performed separately. To identify most suitable time series modeling method the fitted time series models of the each station were compared with time series models of the clustered stations in WL2b region for SPI6 time. It was identified that fitting time series model to forecast extreme drought events in clustered station in WL2b region is more accurate than fitting time series models for individually in each station in this region. It was noted that WL2b region, Horakelle, Palugaswewa stations in IL1 region and Ambalanthota station in DL3 region are drought prone AER in Sri Lanka.