Changing climate, land use change, and invasive species will cause significant impacts to ecosystem structure and function. In order to prepare for these changes, society needs ecological forecasts. Ecological forecasts are quantitative predictions that are critical for documenting and advancing scientific understanding and useful in societal application of knowledge (Katz and Murphy, 2005). Forecasting is necessary for advancing theory because it regularly confronts theory with observations via predictions. Ecological forecasting includes two closely related activities. The first is similar to a weather forecast, that is, an attempt to discern the most likely future state of an ecological system. The second activity adds an additional factor to study the most likely future state of a system, given a decision today (Clark et al., 2001). The first activity is often relevant for short-term forecasts where the system’s own dynamics most strongly govern its change over time (for example, forecasting the likely rate of spread of an invasive species). The second comes into play when alternate management actions or scenarios are being considered (for example, forecasting the likely impacts of alternate forest fire risk mitigation practices on biodiversity). While ecological forecasting typically requires deterministic knowledge of the process being modeled, forecasts are usually probabilistic and provide an estimate of the probability of the future state, not just a point estimate of its value. This session examines ecological forecasts across ecosystems, communities, and populations that are informed by observational data, process studies, and experiments. The session will introduce concepts of forecasting and data assimilation. Speakers will discuss ecosystem forecasts of the carbon cycle in both unmanaged ecosystems and highly managed agricultural settings. The session will also consider forecasts of phenology, community dynamics and populations under changing climate conditions.