Confined pairs of men for 8 days under an experimentally manipulated factorial combination of mission-length expectation, stimulation, and privacy. Subjective reports of stress and anxiety reactions to the confinement situation were obtained via self-report questionnaires administered during the experience. Measures of a trait and a state and subjective stress provided a common picture of the difficulties in isolated small groups. These difficulties were further evidenced by a high abort rate (53%). Groups who expected to remain in isolation a long time but who were unable to complete the mission reported more anxiety and stress than those who expected short missions and were unsuccessful. Additionally, groups under long mission expectations who were successful reported more anxiety and stress than successful groups under short mission expectations. Finally, the austere conditions of a long mission in separate compartments with no outside contact proved to be most stressful. Short missions with or without contact with the outside world were least stressful and anxiety provoking.