Drought is a recurrent climate hazard and it can occur in all types of climates. In this study, recurrence of drought over South Asia has been studied by analyzing observed as well as modeled outputs of meteorological data. The 20th century precipitation has been analyzed over Asia region highlighting the inter-dependence of southwest monsoon and East Asia monsoon focusing on precipitation characteristics over South Asia. Tele-connections of rainfall over south Asia with oceanic indices and the impact on the shifting of the sub-tropical jet have been explored. The study also examined seasonal variability during the 1999-2001 drought periods from the normal years to identify the causes of drought over South Asia. The region receives most of its precipitation (70%) during southwest monsoon season that prevails from June to September. Monsoon season exhibits higher variability compared to winter season as depicted during the analysis of 1901-2010 dataset. Decadal average monsoon rainfall analysis depicts significant (99 %) increasing trend during 1st half of twentieth century (1901-1950). Afterward, significant decreasing trends (95%) are found in the data sets of 1951-2010 and 1981-2010 for South Asia. Winter rainfall is only one eighth of monsoon rainfall. Therefore, the major contribution in terms of water availability comes from southwest monsoon. Hence, failure of monsoon rainfall over the region would lead to drought conditions over South Asia. The analysis of country level precipitation depicts that except for Pakistan, precipitation is exhibiting decreasing trend over Sri Lanka, India and Nepal. However the change is not statistically significant. The correlation between monsoon rainfall and solar index is also examined that depicts that seasonal abnormality increases during steady periods of solar activity and vice versa. Singular Value Decomposition (SVD) analysis is applied to depict that monsoon rainfall over drought vulnerable areas of west South Asia is highly correlated with rainfall activity over monsoon trough located at central India whereas rainfall over eastern Himalaya region is negatively correlated. Significant phases of oceanc indices such as Indian Ocean Dipole (IOD), Pacific Decadal Oscillation (PDO) and Southern Oscillation Index (SOI) have also been identified by computing standard deviation of seasonally averaged index values and investigated their impact on South Asia monsoon rainfall. Analysis depicts that during the past 60 years, frequency of negative episodes (< -1 St dev.) of PDO and SOI have significantly increased. Observed negative phases are twice as compared to positive phases since 1950. Whenever, the two indices are in opposite phase (1997, 2010), it resulted in extreme precipitation episodes over northwest domain of South Asia comprising northern India and Pakistan causing severe flooding over the region. Analysis reveals that IOD significantly contributes in regulating rainfall activity over South Asia. When IOD negative phase persists for longer duration and continue into next year, it weakens monsoon currents resulting suppressed rainfall activity over western India and most parts of Pakistan leading to drought. Analysis of the impact of various indices on case to case basis depicts that negative phase of SOI and positive phase of IOD significantly contribute in suppressing monsoon rainfall activity over South Asia that lead to drought events in the region. The analysis of area-weighted mean seasonal and annual rainfall for southern arid region of Pakistan has revealed that lower terciles are increasing with time. The seasonal area weighted average rainfall for this region is 58.6 mm, whereas the standard deviation is 30 mm, depicting that monsoon rainfall variability in the region is high. The annual and seasonal rainfall in the region has shown negative trend with increased frequency of lower terciles, making the region highly vulnerable to droughts. The hindcasts of coupled models under the EUROSIP project have been analyzed to develop drought prediction tool for the arid regions of Pakistan. The multi-model ensembles have been investigated for the prediction of two categories; below the median and lower-tercile (dry-season). The forecasts performance have been evaluated by employing Relative Operating Curve (ROC) verification method and areas under the curve are also verified through cross validation. The results for deficient monsoon rainfall predictability using the EUROSIP multi-model ensemble system have established that the system reveals good skill for lower-tercile forecasts over southern Pakistan (skill > 0.7). Therefore it is concluded that climate models of European Centre for Medium Range Forecast (ECMWF), Météo-France, and United Kingdom Meteorological Office (UKMO) are sensitive enough towards precipitation formation mechanism over monsoon region and able to predict dry episodes of the climate over drought prone areas of Pakistan.