During the 1980s the number of institutional beds in Sweden has decreased in relation to the number of elderly in the population. In some communities this development has been drastic. The present paper describes how the implementation of this process in one of these communities was based on action research design. Initially most people regarded the development as a success: more elderly people were able to stay in their homes and institutional places were used more effectively. The situation today is more complicated. Elderly patients finally ending up in institutions are in greater need of care than before. Over-occupation of beds is again common, making the situation more "heavy-going" for the staff. Staff outside the institutions are also experiencing a heavier workload. In view of this, the wisdom of further de-institutionalization is now questioned.