The inmates had been randomly assigned to one of two jail units that were identical in design and management, but varied in population density levels. Each unit included 48 rooms (6 tiers, each with a cluster of 8 rooms) and had television, dining, and recreational facilities. Initially population density was rated at capacity on one unit, and above capacity on the other. Data were collected before and after a court order that caused population levels to be decreased on the unit over capacity and increased on the other, so that after the change density on both units was the same. Comparison data were also collected on several other units. The results revealed both absolute and contrast effects for density levels, but also were affected by the direction of shift. That is, when population levels were equal after the change, perceived crowding and sick call rates were higher on the unit that had experienced an increase in density. Also, as population levels increased, there tended to be an increased in isolated passive behavior and a decrease in isolated active behavior in public areas. Implications for crowding research and for jail design and management are discussed. (Author abstract modified).