Econometric modelling of decision uncertainty has received extensive attention in the contingent valuation literature, but these methods are not directly transferable to the realm of multi-attribute stated preference studies. In this paper, an integrated choice and latent variable model tracing the impact of decision uncertainty on the valuation of flood risk reductions in the Netherlands is developed. The proposed model structure is not subject to the potential endogeneity bias and measurement error issues associated with most applied methods. The driving factors of decision uncertainty are identified through stated choices and a set of self-reported decision uncertainty follow-up questions. The model simultaneously accounts for the impact of decision uncertainty on individual choices and welfare estimates. In the presented case study uncertain respondents are found to make more random choices and select the opt out option more often. Willingness-to-pay for flood risk reductions increases after accounting for these behavioural responses to decision uncertainty.