Trend GDP and output gaps play an important role in fiscal and monetary policy formulation, often including the need for forecasts. In this article, we focus on forecasting trend GDP and output gaps with Beveridge-Nelson trend-cycle decompositions trend-cycle decompositions and investigate how these are affected by assumptions concerning correlated innovations and structural breaks. We evaluate expanding window, one-step-ahead forecasts indirectly for the G-7 countries on the basis of real GDP growth rate forecasts. We find that correlated innovations affect real GDP growth rate forecasts positively, while allowing for structural breaks works for some countries but not for all. In the face of uncertainty, the evidence supports that in making forecasts of trends and output gap policy-makers should focus on allowing for the correlation of shocks as an order of priority higher than unknown structural breaks.