Tropical cyclones (TCs) are the most devastating weather phenomena that trigger massive loss of property and life in the coastal areas of the Bay of Bengal (BoB). Scienti c understanding of TCs occurrence can aid policy-makers and residents in coastal areas to take necessary actions and appropriate planning in advance. In this study, we aimed to examine the possible linkage of weather parameters with deadly 22 TCs events in the BoB from 1975 to 2014 using principal component analysis, K-mean clustering and General circulation model (GCMs). Results showed that among 22 TCs, cluster 1 belongs to 12 TCs which occurred under the same atmospheric situation when the sea level pressure (SLP) was below 990hPa, and the temperature ranged from 300C to 390C. A deep negative anomaly of SLP and temperature was observed up to 500hPa levels. In contrast, a negative depression was found at 300hPa geopotential height (GPH) over the study area. Cluster 2 consisted of 9 TCs when SLP was below 1000hPa, and the average temperature was 33.50C. A strong negative anomaly was noticed when surface level up to 500 hPa GPH, but dramatically this depression was completely absent at 300hPa geopotential height over the BoB and entire coastal region. Cluster 3 contained only 1 TCs when the atmospheric circumstance was completely diverse, and the SLP was above 1000hPa. The results of the GCM model revealed that the SLP was lower, and the temperature was higher over BoB compared to the North Indian Ocean. We identi ed the larger depression of SLP and unpredictable temperature anomalies at the upper atmosphere that can trigger an enormous unpredictability throughout the atmospheric level, leading to severe TCs. The outcomes of this study can improve our understanding of weather variables in the upper atmospheric column for forecasting the TCs system more accurately in the future.