PurposeWe investigate causes for the cost overrun and delay of the railway project Stuttgart 21. Besides, we try to forecast the actual costs and completion date at an early stage.Design/methodology/approachThe results of exploratory research show the causes for the cost overrun and delay of Stuttgart 21; we compare our findings with other railway projects. To estimate the costs at an early stage, the reference class forecasting (RCF) model is applied; to estimate the time, we apply an OLS regression.FindingsWe find that the following causes are relevant for the cost overrun and delay of Stuttgart 21: scope changes, geological conditions, high risk-taking propensity, extended implementation, price overshoot, conflict of interests and lack of citizens' participation. The current estimated costs are within our 95% confidence interval based on RCF; our time forecast underestimates or substantially overestimates the duration actually required.Research limitations/implicationsA limitation of our approach is the low number of comparable projects which are available.Practical implicationsThe use of hyperbolic function or stepwise exponential discount function can help to give a clearer picture of the costs and benefits. The straightforward use of the RFC for costs and OLS for time should motivate more decision-makers to estimate the actual costs and time which are necessary in the light of the rising demand for democratic participation amongst citizens.Social implicationsMore realistic estimates can help to reduce the significant distortion at the beginning of infrastructure projects.Originality/valueWe are among the first who use the RCF to estimate the costs in Germany. Furthermore, the hyperbolic discounting function is added as a further theoretical explanation for cost underestimation.