This study proposes to test the respondents' rationality hypothesis as usually assumed by stated preference valuation methods. It focuses on Choice Experiment. To achieve this goal, we implement some rationality tests provided by the Choice Modelling literature. The survey concerns forest recreation. We look at the existence of lexicographic preferences, at the influence of consistency and at the role of task complexity in the individual choice process. We improve the design of the consistency test by varying the rank of the initial and of the repeated choice sets across the versions of the questionnaire. Tests and econometric results show that the choice sets orderings do not impact on choice probability. If a violation of continuity or consistency influences this probability, the Willingness-To-Pay estimates calculated using the total sample and the sub-samples of lexicographic preferers or unstable choosers are not significantly different.