This paper proposes a framework to produce multi-horizon forecasts of business cycle turning points, average forecasts of economic activity as well as conditional forecasts that depend on whether the horizon of interest belongs to a recession episode or not. Our forecasting models take the form of an autoregression (AR) of order one that is augmented with either a probability of recession or an inverse Mills ratio. Our empirical results suggest that a static Probit model that uses only the TS as regressor provides comparable fit to the data as more sophisticated non-static Probit models. We also find that the dynamic patterns of the term structure of recession probabilities are quite informative about business cycle turning points. Our most parsimonious AAR model delivers better out-of-sample forecasts of GDP growth than the benchmark models considered. We construct term structures of recession probabilities since the last oficial NBER turning point. The results suggest that there has been no harbinger of a recession for the US economy since 2010Q4 and that there is none to fear at least until 2018Q1. GDP growth is expected to rise steadily between 2016Q3 and 2018Q1 in the range [2.5%,3.5%].