There are a multiple definitions of resilience for different situations and from different scientific disciplines. They all have at their core the ability of a system to react and respond effectively to threats or disturbance. Adaptation to external drivers and pressures is a critical aspect of resilience.  Mathematical models can be used to measure and analyse some types of resilience but it is a field in which what you can measure is not always important and what is important cannot always be measured.   As the science of resilience develops to provide an alternative and/or complement to more traditional risk assessment approaches in areas such as disaster management, it is also timely to turn the spotlight the other way and consider the resilience of science systems themselves.  Many would argue that science is under threat and needs to rapidly adapt to a number of important external drivers. Indeed, some would argue that science is in crisis in a post-truth world, where scientific 'breakthroughs' are increasingly over-hyped and  under-substantiated.  So how can our science systems adapt and change to better cope with the pressures and demands that we put on them? How can we build resilience into science as an essential pre-condition for science to function effectively in the service of society?