Soil moisture plays an important role in partitioning rainfall into runoff and evapotranspiration. Due to advancements in remotely sensed soil moisture data acquisition techniques, many soil moisture data assimilation (SMDA) studies have been conducted to improve streamflow prediction. It is thus expected that the outcome of a SMDA exercise will be determined by the quality of soil moisture data in hand and the hydrology of the catchment. Our study begins with the hypothesis that it is possible to determine the usefulness of a satellite-based soil moisture data product for areas where paramaterizing a complex model is difficult due to availability of limited information. To this end, we use satellite-based GLDAS root-zone soil moisture data with dynamic Budyko (DB), rainfall-runoff model, to improve streamflow prediction for 60 US basins. Our results suggest that there is a reasonably good one-to-one or universal relationship between instantaneous dryness-index (φ), the key state variable of the DB model, and root-zone soil moisture (θ), which implies that the model can directly use soil moisture information for predicting streamflow. To check the robustness of the universal φ-θ relationship, we also developed basin specific φ-θ relationships. Model performance, expressed in terms of Nash-Sutcliffe efficiency (NSE), improved in 34 basins when we considered the universal φ-θ relationship and in 51 basins when we considered the basin specific relationships. Multiple linear regression (MLR) analysis reveals that change in NSE due to the use of soil moisture information is predicted quite well by certain basin characteristics. In particular, it is found that available water capacity and forest area positively influence NSE, whereas average sand, silt and clay contents, latitude, and longitude affect it negatively. A slightly stronger MLR relationship was observed while considering soil moisture deficit (the difference between saturated soil moisture and actual soil moisture) in place of soil moisture. From the stepwise MLR analysis, it is observed that the vegetation and runoff generation mechanism control the parameters of the φ-θ relationship. Overall, our study provides a framework to determine the suitability of remotely-based soil moisture data for hydrological modelling. © 2020 Elsevier B.V.