There is an increasing interest in the role of individuals’ rates of time preference in the decision to invest in health. However, there is no consensus on how best to elicit time preference and relatively few studies have tested the robustness of elicitation methods. This potential for measurement error may help explain why the correlation between time preference and health behaviours tends to be relatively small even though time preference is a key parameter in economic models. The aim of this paper is to test the robustness of a popular method for eliciting time preference (multiple price list). We test for order effects and examine the impact of a truth telling oath. We conduct exploratory analyses to examine the predictive validity of the different time preference estimates for smoking behaviour. We find strong order effects highlighting the importance of understanding design features when eliciting and interpreting experimental measures of time preferences