The welfare of animals in extensive systems has received little attention despite significant welfare challenges in these environments. Additionally, recent reform of Common Agricultural Policy payments in the EU have put increasing financial emphasis on farmers to attain improved levels of animal welfare, although suitable methodologies for assessing animal welfare are lacking. Assessment of welfare in intensive systems frequently involves assessing compliance with buildings and space requirements, as well as behavioural observations of animals in their home pens and monitoring of health records. In extensive systems, however, many of these measures are inappropriate or impossible. Environments are often heterogeneous, animals may be difficult to observe, individuals may not be identifiable and health records may be limited or completely lacking. The expression of fear by extensive animals, which rarely come into close contact with humans, often limits the value and practicality of behavioural observations as welfare indicators. Currently, welfare audits of extensive farms rely greatly on mortality and morbidity records to determine welfare status. Whilst these measures are important, the strong health monitoring bias does not adequately allow an interpretation of the emotional experiences of extensive animals. We suggest that developing methodologies based on assessing key features of the environment, such as handling facilities, assessment of the skills, knowledge and planning of the stockperson, and assessing the animal at key points in the production cycle when they are gathered may allow a workable methodology for assessing animal welfare to be developed for extensive systems.