The social environment is very important for the welfare of animals in loose housing dairy production systems. This article reviews recent literature on the effect of animal density (AD) and regrouping (RG'j on the welfare of cattle and describes the development of feasible indicators for the social environment. Special emphasis is given to the methodological problems that arise when AD and RG are used as welfare indicators in a welfare assessment at the herd level. Various factors affecting estimates of AD were considered, including the size of the animals, correction for very high AD values, pen shape and how best to aggregate the results at herd level and over time. The examination of RG is centred around the effect of early social experience of the animals, the stability of social relationships, and the effect of pen changes.

A range of parameters is suggested for the evaluation of AD and RG as possible welfare indicators. These are based on observational data from 10 Danish dairy herds and related to clinical records from the herd farms. It is concluded that mean AD is not feasible as a welfare indicator at the herd level but the 25th percentile of AD corrected for the liveweight of the animals should be used instead. The two most promising parameters for evaluation of RG are the frequency of combined pen and group changes for a sample of the herd, and the probability of a certain duration of inter-animal relationships. Results from clinical observations correlated with neither AD nor RG.