On the presumption that evaluation of the effectiveness of policy is a necessary precursor to policy change, effectiveness is usually evaluated empirically in complex settings; although in some cases violation of ‘natural laws’ may also be used. In the case of flood mitigation policy, and all imaginable cases of environmental disaster mitigation, empiricism is required that involves statements about causality. Some philosophers of science have argued that causality and counterfactuals are linked. David Lewis’ famous formulation is as follows: c is a cause of e if it is the case that for two actual events c and e, had c not occurred, e would not have occurred. This formulation can be seen to be useful in the case of clinical trials: but can it be applied in non-experimental settings? For example, if embankments had not been built then either flood damage would not have declined or it would have been much worse. In the Brahmaputra River the simple, and probably simplistic question: have embankments reduced flood damage? can be answered but the effect of no embankments can only be inferred, suggesting another question: is the evidence for the effectiveness of embankments sufficient to justify their continuation?