Human serum lipoproteins are known to participate in or modify several immunologically relevant responses, including the inhibition of target cell lysis initiated by fluid-phase C5b-7 (reactive lysis). We now report that human high density lipoproteins (HDL) can inhibit the complement (C) lytic mechanism after C5b-7, C5b-8, and even C5b-9 have been bound to the target membrane. This inhibitory activity of serum or plasma copurifies in hydrophobic chromatography with antigenically detected apolipoprotein A-I (apoA-I), the major HDL apoprotein, and with HDL in CsCl density gradient ultracentrifugation. Although HDL is more active than its apoproteins in fluid-phase inhibition of C5b-7-initiated reactive lysis, the HDL apoproteins are more effective after C5b-7, C5b-8, or C5b-9 have become bound to human or sheep erythrocytes (E). Highly purified HDL apoproteins, apoA-I and apoA-II, both have greater inhibitory activity than whole HDL on a protein weight basis, and some evidence has been obtained that apoA-I dissociating spontaneously from HDL may be the principal inhibitory moiety in physiological situations. HDL lipids themselves are inactive. The HDL-related inhibitors are ineffective when incubated with EC5b-7 and removed before C8 and C9 are added, and only minimally effective on cell-bound C5b-8 sites before C9 is added. They exert their most prominent inhibitory activity after C9 has been bound to EC5b-8 at low temperature, but before the final temperature-dependent, Zn(++)-inhibitable membrane damage steps have occurred. Therefore, HDL or its apoproteins do not act to repair already established transmembrane channels, but might interfere either with insertion of C9 into the lipid bilayer or with polymerization of C9 at C5b-8 sites. This heat-stable inhibitory activity can be demonstrated to modify lysis of erythrocytes in whole serum, i.e., it does not depend upon artificial interruption of the complement membrane attack sequence at any of the above-mentioned stages. Contributions of the target membrane itself to the mechanism of inhibition are suggested by the observations that, in contrast to sheep or normal human E, lysis of guinea pig E or human E from patients with paroxysmal nocturnal hemoglobinuria is inhibited poorly. This is the first description of a naturally occurring plasma inhibitor acting on the terminal, membrane-associated events in complement lysis. Although further study is required to assess the physiologic or immunopathologic significance of this new function of HDL, the HDL apoproteins or their relevant fragments should be useful experimentally as molecular probes of the lytic mechanism.