Contrary to earlier reports, we have found that tri- and hexapeptides analogous or homologous with segments of the 23-residue N-terminal fusion sequence (FS) of the viral transmembrane glycoprotein gp41 (residues 517-539) did not significantly inhibit HIV-1-induced syncytium formation, using an uninfected cell-infected cell fusion assay. In contrast, we found that the high molecular weight apolipoprotein A-1 and a 23-residue analog of the FS, with the phenylalanine residues at positions 524 and 527 replaced with alanine residues, were effective inhibitors. Although the tripeptides were ineffective as inhibitors of syncytium formation, we found a number of them inhibited red cell lysis induced by the synthetic peptide AVGIGALFLGFLGAAGSTMGARS (based on the HIV-1 gp41 FS). This effect was also seen with apolipoprotein A-1. The Ala524,527 analog of the fusion sequence could not be tested in this system because it was hemolytic. We concluded that the smaller peptides were effective inhibitors of hemolysis because they interfered with pore formation by the fusion sequence peptide, either by disrupting the pores or by preventing the peptide from adopting the alpha-helical conformation found in the pores. On the other hand, membrane fusion, which is a prelude to syncytium formation, has been shown to require the fusion sequence in the beta-strand conformation. We argue that small peptides would be unable to block interaction between such strands, although larger molecules, such as apolipoprotein A-1 and the Ala524,527 analog, would be able to do so and thus inhibit fusion. It seems, therefore, that a successful drug directed against the FS-cell membrane interaction stage of syncytium formation would need to be of relatively high molecular weight and complexity.