In this document, I present an overview of my research in hydrology over the last ten years and the directions I would like to follow in the future. In the first chapter, I review the hydrological literature in order to place the approach I adopted (together with the rest of the research team with whom I have been working since 1995) in the wider context of hydrological modeling approaches. My preferred method, which I would define as a downward empirical approach, appears quite original compared to the mainstream of hydrological research, especially in its treatment of time and space. Also, I show how it can shed new light on the old passionate debates that have flourished within the hydrological community over the last forty years. The second chapter is devoted to a presentation of my main published research results. It is organized so as to show the value of assessing the sensitivity of watershed models to their inputs. I cover successively model sensitivity to the input of: precipitation, discharge, potential evapotranspiration, and distributed rainfall. Then, I present a synthetic diagnosis of the main scientific problems, which, I believe, plague modern hydrological models and impede future advances, on both the applied and theoretical sides of hydrology. The third chapter gives a presentation of my future research objectives, organized into three groups: applied hydrological questions (with the issues of flood and drought forecasting and snow hydrology), theoretical hydrological questions (with the issues of ensemble modeling, ungaged basins and the study of natural hydrological behavior variability) and last, interesting but still unripe hydrological questions (prediction of the impact of land-use change and spatialized hydrological modeling).