Abstract: Research on the atmospheric predictability and causes of forecast errors has important significance to improvements of numerical weather prediction and operational forecast skill. Based on flow-dependent ensemble, ensemble sensitivity determines relationships between forecasts and initial or early forecast atmospheric conditions. It provides a new methodology to reveal dynamic features relevant to the predictability of the forecast variable of interest and to understand origin of forecast errors with its propagation. This paper introduces the definition and measures of ensemble sensitivity and then reviews its application progress for typical weather systems and high-impact weather. Finally, the strength and limitation of the ensemble sensitivity is discussed.