A double-blind study of the effects of phlebotomy was carried out in 18 patients with polycythemia secondary to severe hypoxemic lung disease. Eleven subjects underwent a single phlebotomy of 10 percent of their blood volume, and eight patients serving as controls underwent a sham procedure. Eight of the phlebotomized subjects, but none of the controls, reported subjective clinical improvement (P less than 0.005). Subjects who noted improvement after venesection had higher hematocrit readings than those who did not (P less than 0.02). Symptomatic relief seemed to be most dramatic in those with clinical evidence of congestive heart failure. In contrast to this clear-cut subjective improvement, phlebotomy did not alter objective indices of airway obstruction, lung elastic recoil, pulmonary gas exchange, or exercise tolerance in either the phlebotomized or the control group. Thus, although phlebotomy produced subjective benefit in the majority of patients studied, it was not associated with objective improvement in lung function or exercise tolerance.