A better understanding of how society anticipates and adapts to future changes is critical to inform impact assessment and to develop timely and well-targeted policies to support adaptation. However, the forward-looking adaptation process remains poorly understood. In this paper we introduce choice experiment as a useful approach to investigate how households prefer to adapt livelihoods ex ante to the economic impact of climate and policy changes. This allows us to frame adaptation decisions within the random utility theory and explicitly quantify the likelihoods of particular adaptation choices given varied attributes of contextual changes and households. We collected data from 162 rural households in three Chinese mountain villages. Overall, households chose primarily to increase efforts in agriculture activities or stick to current livelihood portfolios. The results of a Mixed Logit model indicated that households’ choice of agriculture was certain while their adoption of non-agriculture liveliho ods to safeguard the households from future changes. Moreover, several possibilities were evaluated for policy interventions to build adaptive capacity of households and facilitate adaptation. Such measures could, for instance, focus on supporting agricultural inputs, providing access to credit as well as practical skills training.