The human disease paroxysmal nocturnal hemoglobinuria (PNH)' may be regarded as an experiment of nature that has already provided, and should continue to provide, unique insights on the cytolytic function of complement (C) and its regulation (1-4). The studies of our laboratory have been focused particularly on elucidating the process by which the most sensitive type of PNH erythrocyte (E) (designated type III) are abnormally sensitive to in vitro reactive lysis, i.e., to membrane attack initiated by purified C5b6, C7, C8, and C9 in the absence of the earlier activation and amplification steps (5).In an earlier report (6), we presented evidence that for a given input of purified C5b6, type III PNH E (PNH-III E) did not bind '21 I-C7 to a greater extent than normal human E. Rather, the data indicated that PNH E underwent two-fourfold greater lysis for a given number of C7 molecules bound. Compatible results were obtained in less extensive, unpublished studies with radiolabeled C5b6. Compelling independent evidence from other laboratories (7) indicates that the basic unit of the membrane attack complex involves one molecule each of C5b, C6, C7, and C8 in the formation of the C5b-8 complex, to which a variable number of C9 molecules then bind to produce the fully effective membrane attack complexes. Therefore, although C8 uptake has not been directly measured in our system, it seems justified to infer that in reactive lysis the assembly of C5b-8 complexes on PNH-III E from the fluid phase is quantitatively normal . The excessive lysis of the PNH E by the full C5b-9 sequence could result from a quantitative abnormality affecting C9 binding and/or polymerization within the C5b-9 complex, or from a qualitative difference in the disposition or function of the terminal C components after they have become bound to the PNH membrane. The previously reported deficiency of a membrane complement regula-