In a retrospective study of laboratory tests for paroxysmal nocturnal hemoglobinuria (PNH), the red blood cells of several patients gave conflicting results when different testing procedures were used. To resolve these discrepancies, an examination of the various procedures was undertaken, using sulfhydryl reagent-treated red blood cells as model PNH cells. Rabbit antiserum to human red blood cells was used to sensitize erythrocytes for specific antibody-dependent complement lysis sensitivity tests. Acidified serum and inulin were used in fluid-phase complement activation tests. For the general laboratory, a modification of the sucrose lysis test is described, which can give the relative complement sensitivity and size of an abnormal subpopulation of red blood cells. Considering the results of our patients and the reports in the literature of the association of PNH with a variety of lymphoproliferative and myeloproliferative disorders, it is stressed that testing procedures for PNH be controlled more carefully for a better understanding of the nature of the membrane defect.