To determine whether a moral reasoning exercise can improve response quality to surveys of healthcare priorities A randomised internet survey focussing on patient age in healthcare allocation was repeated twice. From 2574 internet panel members from the USA and Canada, 2020 (79%) completed the baseline survey and 1247 (62%) completed the follow-up. We elicited respondent preferences for age via five allocation scenarios. In each scenario, a hypothetical health planner made a decision to fund one of two programmes identical except for average patient age (35 vs 65 years). Half of the respondents (intervention group) were randomly assigned to receive an additional moral reasoning exercise. Responses were elicited again 7 weeks later. Numerical scores ranging from -5 (strongest preference for younger patients) to +5 (strongest preference for older patients); 0 indicates no age preference. Response quality was assessed by propensity to choose extreme or neutral values, internal consistency, temporal stability and appeal to prejudicial factors. With the exception of a scenario offering palliative care, respondents preferred offering scarce resources to younger patients in all clinical contexts. This preference for younger patients was weaker in the intervention group. Indicators of response quality favoured the intervention group. Although people generally prefer allocating scarce resources to young patients over older ones, these preferences are significantly reduced when participants are encouraged to reflect carefully on a wide range of moral principles. A moral reasoning exercise is a promising strategy to improve response quality to surveys of healthcare priorities.