In numerical weather prediction, the four-dimensional variational (4DVar) data assimilation (DA) is one of the most operative techniques to provide better initial and lateral boundary conditions and, thus, weather forecasts. However, the 4DVar DA system requires a high cost of running sequential iterations of linear and adjoint models. The four-dimensional ensemble variational (4DEnVar) DA is evolving over the 4DVar, which combines a climatological covariance model with covariances generated from an ensemble of forecasts to sample the current uncertainties. In this study, the weather research and forecasting model was used to evaluate the behaviour of the 4DVar, and 4DEnVar DA approaches in the simulation of two tropical cyclones, Titli and Lehar, over the Bay of Bengal basin of the North Indian Ocean by assimilating conventional and satellite radiances observations. An overall analysis indicates that the 4DEnVar simulations performed relatively better in cyclones’ track prediction and more or less similar to the 4DVar for other cyclone-related characteristics.