In this paper, we seek to identify robust predictors of individuals’ attitudes towards climate change and environmental degradation. While much of the extant literature has been devoted to the individual explanatory potential of individuals’ characteristics, we focus on the extent to which these characteristics provide robust predictions of climate and environmental attitudes. Thereby, we adjudicate the relative predictive power of psychological and sociodemographic characteristics, as well as the predictive power of combinations of these attributes. To do so, we use a popular machine learning technique, Random Forests, on three surveys fielded in China, Switzerland, and the USA, using a variety of outcome variables. We find that a psychological construct, the consideration of future consequences (CFC) scale, performs well in predicting attitudes, across all contexts and better than traditional explanations of climate attitudes, such as income and education. Given recent advances suggesting potential psychological barriers of behavioural change Public (Weaver, Adm Rev 75:806–816, 2015) and the use of psychological constructs to target persuasive messages (Abrahamse et al., J Environ Psychol 265–276, 2007; Hirsh et al., Psychol Sci 23 578–581, 2012), identifying important predictors, such as the CFC may allow to better understand public’s appetite for climate and environmental policies and increase demand for these policies, in an area where existing efforts have shown to be lacking (Bernauer and McGrath, Nat Clim Chang 6 680–683, 2016; Chapman et al., Nat Clim Chang 7 850–852, 2017).