Water resource assessment is important for integrated water resources management of a basin to which rainfall is a vital component. Generally, semi-distributed to distributed models are used for water resource management studies. Higher data requirement and simulation time restrict the use of these models. Lumped conceptual models are drawing attention these days owing to their simplicity and minimum data requirement. In the present study, the performance of conceptual models Genie Rural a 4 parametres Journalie (GR4J) and Australian Water Balance Model (AWBM) was evaluated in comparison with semi-distributed Soil and Water Analysis Tool (SWAT) model. The selected study area, Upper Godavari River Basin, is in the windward side of Western Ghats in India, which receives heavy rainfall during south-west monsoon season. However, this region faces water scarcity issues in non-monsoon period due to lack of proper water management scenarios. Five catchments in upper Godavari basin are used for the analysis. Spatiotemporal analysis of rainfall is done to understand its effect on streamflow at the catchment outlet. The efficacy of model predictions has been analysed statistically in terms of Nash–Sutcliffe Efficiency, coefficient of determination (R2) and percentage bias. Flow duration curves and time series diagram of streamflow predictions are also compared to better understand the results. All three models predicted streamflow with reasonable accuracy. Considering structural simplicity, less data requirement and simplicity in calibration process, this study proposes conceptual models over SWAT in the regions facing data scarcity.