Current scientific progress in the understanding of animal welfere is resulting in the possibility of realistic assessments of welfare and the application of such assessments. Understanding has been advanced by the recognition that welfare is not a unitary variable but has different aspects, which may be compromised independently. Some problems which arise, such as injury and disease, also reduce production and may be covered by veterinary and production science. Others, though, such as behavioural restriction and stress, do not affect production so clearly. There is increasing input to these areas from behavioural science and physiology. Ethology has contributed theoretical insights into the control of behaviour, basic knowledge of the behaviour of farm animals and comparisons between different environments (e.g. intensive, extensive and feral). It has also highlighted occurrence of abnormal behaviour, such as stereotypies and cannibalism, in some circumstances. To investigate causes of abnormal behaviour, though, observation must be complemented by experimentation, and by behavioural physiology and neurophysiology. These are now producing notable advances in our understanding of stress. Experimental work is also progressing on motivation for normal behaviour, such as feeding,and for different features in the environment. With increased understanding of different aspects of welfare comes a move away from unrealistic expectations of unitary welfare meesurements or indices, and towards an acceptance that there are scientific grounds for abroader assessment. Further, there is increasing recognition that assessment of welfare has a valid, scientific role in the design end management of housing and husbandry systems for our farm animals.