Based on metatheoretical considerations, this article discusses what kinds of traffic forecasts are possible and what kinds are impossible to make with any reasonable degree of accuracy. It will be argued on ontological and epistemological grounds that it is inherently impossible to make exact predictions about the magnitude of the ‘general’ traffic growth 20-30 years ahead, since many of the influencing factors depend on inherently unpredictable geopolitical trajectories as well as contested political decision-making. Due to the context-dependency of each particular (...) planning situation, it is also hardly possible to make exact, quantitative predictions about the impact of implementing a specific infrastructure project, compared to ‘doing nothing’. Instead of relying on traffic model simulations as the general forecasting and assessment tool in transport planning, we propose to separate the so-called strategic, tactical and operational levels of traffic forecasting into three distinct methodological approaches reflecting the different degrees of openness/closure of the systems at hand: scenario analyses at the strategic level; theory-informed, mainly qualitative analyses supplemented with simple calculations at the tactical level; while more traditional micro-simulations should be applied only at a detailed operational level. (shrink)