In this study we tested the hypothesis that fusion mediated by the influenza virus hemagglutinin (HA) is a cooperative event. To so this we characterized 3T3 cell lines that express HA at nine different defined surface densities. HA densities ranged from 1.0 to 12.6 x 10(3) HA trimers/microns2 as determined by quantitative fluorescent antibody binding. The lateral mobility and percent mobile fraction of HA did not vary significantly among these cells, nor did the contact area between HA-expressing cells and target RBCs. The fusion reaction of each HA-expressing cell line was analyzed using a fluorescence dequenching assay that uses octadecylrhodamine (R18)-labeled RBCs. For each cell line we measured the lag time preceding the onset of fusion, the initial rate of fusion, and final extent of fusion. The final extent of fusion was similar for all cell lines, and the initial rate of fusion as a function of HA surface density displayed a Michaelis-Menten-type dependence. However, the dependence of the lag time preceding the onset of fusion on HA surface density was clearly sigmoidal. Kinetic analysis of the data for the reciprocal lag time vs HA surface density, by both a log/log plot and a Hill plot, suggested that the observed sigmoidicity does not reflect cooperativity at the level of formation of HA aggregates as a prerequisite to fusion. Rather, the cooperativity of the process(es) that occur(s) during the lag time arises at a later step and involves a minimum of three, and most likely four, HA trimers. A model is proposed to explain HA cooperativity during fusion.