Animal welfare in livestock systems is strongly dependent on husbandry conditions. ‘Comfort Class’ is defined as a specific minimal level of husbandry conditions of animals, at which the ability of animals to meet their needs is not compromised by husbandry conditions. It is assumed that if this level is attained, animal welfare (the quality of life as experienced by the animal) will not be restricted by husbandry conditions. This idea of a Comfort Class level originates from an early interdisciplinary innovation trajectory in pig husbandry and was adopted by the major Dutch pig farmers' organisation and the major Dutch animal protection organisation. These two parties built a Comfort Class facility as a proof of principle to test and demonstrate the idea. The aim of the first study in this facility was to empirically test whether a facility that meets the Comfort Class level results in good animal welfare for pigs. In two batches each with 144 undocked pigs, housed in three group sizes, observations were made on tail integrity, skin lesions, activity patterns and degree of synchronisation and clustering of eating and lying behaviour. In the first batch, tail and skin damage was at a low level. The second batch, starting with 64% of the animals having bitten tails, ended with almost all tails recovered. Activity patterns were similar for group sizes and growing stages. Resting behaviour was highly synchronised, but rather spread out over the lying area. Synchronisation of eating was limited, as 52% of the meals were taken alone and a further 29% by two pigs together. The study offered support for the hypothesis, that the Comfort Class level results in a good quality of life for pigs, especially based on the absence of observed welfare infringements (conclusion 1). Results on space use, synchronisation and clustering indicated that the theoretically derived requirements on space allowance and number of feeders might be reduced without compromising the Comfort Class level. The expected limited statistical power of the experiment did not hinder further development, as, during the process, working on the scientific underpinning of the concept was more important than the actual rigidity of the conclusions. The concurrent scientific activities legitimated the stakeholders' activities and emphasised their claim that practical animal welfare improvement is possible. The project initiated further experimentation and design in practice and contributed to market introduction of welfare improved pork. The methodology applied in the project turned out to be the start of a series of interactive innovation initiatives in animal production sector, leading to the RIO ("Reflexief Interactief Ontwerp", Dutch for Reflexive Interactive Design) innovation approach. Conclusion 2 of this study is that this interactive approach to experimentation facilitates the implementation of science based welfare improvements in practice.