Dysregulated protein kinase activity can cause severe human diseases. Consequently, there is a growing number of kinases that constitute candidate targets for pharmaceutical intervention. However, target validation is critical, since not all kinases are “druggable”, i.e. suitable as a selective drug target. In this review, we briefly introduce major strategies for generating mouse models to analyse and control the expression and function of distinct genes, to confirm specific inhibition of intended drug targets and to identify undesirable secondary effects in vivo. Focussing on tyrosine kinase 2 (Tyk2) and other Janus kinases (Jaks) as case studies we follow the path of investigations from in vitro towards in vivo experimentation. We give examples for the sometimes surprising consequences of the systemic absence of a distinct kinase; consequences that can not be easily deduced solely based on results from in vitro data. We discuss aspects of kinase functions that are distinct from their catalytic activity and give examples for cell-type specific functions. Potential pitfalls (e.g. embryonic lethality, species differences) of using mouse models as experimental systems for studying human diseases are discussed and strategies for improvements to deal with such complications are exemplified.