OBJECTIVES: Using an example of a new drug for rheumatoid arthritis which offers comparable effectiveness and side-effect point estimates to older drugs, we explore preferences for treatments labelled ‘new’. We then examine the persistence of preferences once ambiguity in the evidence base due to it being new is introduced. METHODS: A representative Canadian population sample (n=2837) was randomized to one of three discrete choice experiment (DCE) designs, seeking choices between hypothetical treatments for rheumatoid arthritis with different levels of 7 attributes: route and frequency of administration, chance of benefit, serious and minor side-effects, life expectancy, and uncertainty in benefit and side-effect estimates. DCEs differed in whether the treatment was 1) described as new (recently available) or older (5 or 10 years), 2) whether a qualitative description describing the confidence in the evidence was included instead, or 3) both the length of time available and confidence in evidence was provided. We collected self-reports of respondent innovativeness, numeracy, and risk attitude. RESULTS: Overall, all 6 consistent attributes influenced preferences for treatment. A preference for less ambiguity (more confidence) in benefit and side-effect estimates was observed, but no preference for a treatment labelled ‘new’ or ‘old’. Early adopters (n=173) had a significant preference for ‘newer’ treatments relative to old treatments (B=0.157, p=0.045). The magnitude of preference for new treatments was comparable with preferences for reducing the risks of serious side-effects in this group. When the newness of the drug was combined with ambiguity in the evidence base, these preferences for ‘new’ treatments diminished. CONCLUSIONS: Preferences for innovation in health care exist for some groups of people, but when presented with the implications of new treatments (increased ambiguity in evidence), these preferences diminished. Physicians should either avoid describing whether treatments are ‘new’, or qualify the implications of a ‘new’ treatment in terms of ambiguity in estimates of risks and benefits.