This paper documents a set of models used by Norges Bank in estimating the output gap. The models take into account developments in key cyclical indicators such as GDP, unemployment, inflation, wage growth, investment, house prices and credit growth. As the output gap cannot be observed, there is no direct way of evaluating the estimated output gap. Criteria for a good estimate of the output gap can, however, be the extent to which the output gap estimates provide information about future developments in GDP growth, inflation and unemployment. Measured this way, the models have good forecasting properties compared with simple trend estimations solely based on GDP data. The forecasting properties for an average of the models are shown to be better than for each individual model. The models' estimates of the output gap also have relatively good real-time properties.