The future being uncertain, forecasts ought to be probabilistic in nature, taking the form of probability distributions over future quantities or events. Accordingly, a transdisciplinary transition from point forecasts to probabilistic forecasts is well under way. The ScienceFore project seeks to provide guidance and leadership in this transition, by developing the theoretical foundations of the science of forecasting, as well as cutting-edge statistical methodology, along with applications in meteorology and economics.Theoretically, we will focus on the study of aggregation methods for the combination of multiple probabilistic forecasts for the same quantity or event, and on the design and structure of performance measures that encourage truthful predictions, including but not limited to proper scoring rules. In applications, we will develop statistical postprocessing techniques for the THORPEX Interactive Grand Global Ensemble (TIGGE), which comprises the world's leading global numerical weather prediction models. The key challenge is to retain physically realistic and coherent joint dependence structures across meteorological variables, continents and oceans, and look-ahead times. Furthermore, we will investigate the use of statistical postprocessing techniques in macroeconomic surveys, and aim to resolve a long-standing puzzle in the evaluation of economic and financial forecasts.Theory and applications will intertwine closely, to result in a project that constitutes much more than the sum of its parts. For example, the study of the properties of aggregation methods will inform the development of postprocessing methods for ensemble weather forecasts, and decision theoretically principled approaches to the design of performance measures call for a change of paradigms in the practice of the generation and evaluation of point forecasts, to be demonstrated in case studies.