The scientific community is now developing a new set of scenarios, referred to as Shared Socio-economic Pathways (SSPs) to replace the SRES scenarios. To be used to investigate adaptation and mitigation, SSPs need to be contrasted along two axes: challenges to mitigation, and challenges to adaptation. This paper proposes a methodology to develop SSPs with a “backward” approach. The methodology is based on (i) an a priori identification of potential drivers of mitigation and adaptation challenges; (ii) a modelling exercise to transform these drivers into a large set of scenarios; (iii) an a posteriori selection of a few SSPs among these scenarios, such that they cover the uncertainty space in terms of challenges to adaptation and mitigation. This methodology is applied to the selection of a few SSPs, but it could also be applied to any specific decisions faced by decision-makers. From a large database of runs built by many models, the methodology would allow selecting the most relevant scenarios for a specific decision, i.e. scenarios that best predict when the analyzed choice performs poorly or well.