Over the past decade there has been huge growth in the implementation of animal welfare assessments and audits in a variety of animal applications around the world. These include quality assurance programmes for meat, milk and eggs, certification for specialty brands or food-labelling programmes, accreditation of zoos, laboratories and animal shelters, and proofs of compliance with animal welfare standards required by regulatory agencies. Assessing animal welfare in such practical settings poses challenges at many levels. The animal welfare measures chosen for an assessment must be valid, repeatable and robust, and the sampling techniques must provide accurate representations of the overall welfare status of large groups of animals. Animal welfare assessors can come from a range of backgrounds with varying skill levels and experience and need to receive adequate training to ensure reliability. Although automated measures for animal welfare assessments can reduce potential for assessor error and bias, save time and reduce costs they must be sufficiently validated.