Climate change knowledge can inform regional and local adaptation decisions. However, estimates of future climate are uncertain and methods for assessing uncertainties typically rely on the results of climate model simulations, which are constrained by the quality of assumptions used in model experiments and the limitations of available models. To strengthen scientific knowledge for climate services and climate change adaptation decisions, we explore the use of structured expert elicitation to assess future regional climate change. Using the Lower Yangtze region in China as a case study, we elicit judgements from six experts on future changes in temperature and precipitation as well as uncertainty sources, and compare it with climate model outputs from the Couple Model Intercomparison Project phase 5 (CMIP5). We find high consensus amongst experts that the Lower Yangtze region will be warmer in the coming decades, albeit with differences in the magnitude of change. There is less consensus about the direction and magnitude of future precipitation change. Compared with CMIP5 climate model outputs, experts provide similar or narrower uncertainty ranges for temperature change and very different uncertainty ranges for precipitation. Experts considered additional factors (e.g. model credibility, observations, theory and paleo-climatic evidence) and uncertainties not usually represented in conventional modelling approaches. We argue that, in context of regional climate information provision, expert-elicited judgements can characterise less predictable, or less explored, elements of the climate system and expert elicited reasoning provides additional information and knowledge that is absent from modelling approaches. We discuss the value in bringing together multiple lines of evidence, arguing that expert elicited information can complement model information to strengthen regional climate change knowledge and help in building dialogue between climate experts and regional stakeholders, as part of a more complete climate service.