Over the past decades, ecosystem and climate change modeling have made great advances. However, as indicated by the uncomfortably large deviations between predicted climate change and today's observed effects, there is obviously a lot of room for improvement. Today it is accepted that global warming, and a series of derived effects, are ocurring at much faster rates than predicted even with the most sophisticated models just 10 years ago. Similar discrepancies exist for ecosystem models which try to predict production or the development of pest and disease populations. In the end, the complexities of non-linear behavior and multiple synergistic effects may render such complex systems impossible to model within the limits of acceptable accuracy. If models only hold under extensive lists of unrealistic assumptions (such as linear and additive effects vs non-linear synergistic effects), then their value for deriving practical recommendations must be questioned. So: what are the limits to (meaningful) modeling? I would warmly welcome pointers towards a readable account of this issue.