The present research was an attempt to see how the existing system of commercial group houses with ESF can be made more welfare friendly by modifying the group size and structure. The groups studied showed no difference in terms of major welfare indicators such as cortisol concentrations and number of total aggressive interactions. However, total injury scores were higher and number of non-agonistic social interactions was lower in the dynamic group suggesting that the welfare of sows in the dynamic group was compromised compared to the other groups. The production performance of sows was also similar among the groups. Although, subgroup formation was visible in dynamic group it was apparent only in lying behavior during the period of the study. However, free intermingling of sows might have happened later during their stay in the group. The time required for routine management was higher in the dynamic group. As observed in the present study and in many previous studies, aggression at mixing and competition for feeder entry are the major threats to the welfare of sows in group systems with ESF. It may not be possible to improve the system without addressing these issues. Changing the group size and structure alone may not bring in better welfare. Pen and feeder designs are crucial along with measures to reduce aggression at mixing and hunger. Ensuring control of access to the feeder at the gate rather than only at the point of feed delivery may reduce the aggression associated with feeder entry. Similarly, providing manipulable materials such as a small quantity of straw may help to divert the attention of sows at mixing and may reduce hunger to some extent. Pre-exposure of sows by maintaining sub-groups of sows in adjacent pens may reduce aggression at mixing.