Kites have seen substantial development in the last 10 years, going from mostly toys to high-performance sports-related equipment for e.g. kite surfing. The design process of these kites, however, is mostly a trial-and-error affair. Ten years after the sports-related kite revolution a new development is is emerging: Industrial applications for kites. Systems to propel ships and generate sustainable energy are now under development worldwide at over 40 companies and institutes. These new industrial applications will put strickter and more complex requirements on these kites. The current trial-and-error approach to the design of kites will not suffice. In this thesis a different design methodology is proposed. This methodology leans on three pillars. The first pillar is "Knowledge". As it turns out, there is still a lot unknown about the behavior of kites. This thesis further develops that knowledge. The seccond pillar is "Simulation". Nowadays, a large number of prototypes are produced and tested. So many in fact, that many designers do not even have the time to test them all. With the advance of complex industrial kites, this situation is expected to escalate. The capability of virtually testing kites will shrink the prototype phase into more managable proportions. Furthermore, it will contribute to the understanding of kites as well. This thesis proposes a number of models to simulate kites on a conventional desktop computer. These models include both rigid-body and multi-body models. The latter is capable of simulating a kite including its extreme flexibility. The third and last pillar is "Measurement". Controlled and reproducable measurements are essential for validation and evaluation. The thesis closes with a number of case studies which show the advantages and opportunities of this methodology.