This study examines monthly, seasonal and annual temporal rainfall trends and variability using average monthly time series over the three severe drought-prone Beed, Osmanabad and Latur districts of the Marathwada region of Maharashtra, India (1901 to 2018 for 118 years). Rainfall Anomaly Index (RAI) is used to identify dry and rainy years, and the Coefficient of Variations (CV) is employed to compute rainfall variability. The non-parametric Mann-Kendall (MK) test and Sen’s slope estimator are calculated to detect the monthly, seasonal, and annual significance of trends and the magnitude. The RAI finds many drought and rainy years over the region during the study period. The analysis results show that out of 12 months, a few months observe statistically significant rainfall trends, while some of them observe statically negative rainfall trends at 1 and 5% significance levels. The Winter season records the highest rainfall variability in the four seasons over three districts. The southwest monsoon and annual rainfall series have shown the lowest variability in the data series. Drought (low rainfall) severely impacts on rural livelihoods and agricultural systems, particularly in vulnerable areas where poverty is prevalent, irrigation facilities are scarce and poor productivity.