The shape of lipids has long been suspected to be a critical determinant for the control of membrane fusion. To experimentally test this assertion, we used conical and malleable lipids and measured their influence on the fusion kinetics. We found that, as previously suspected, both types of lipids accelerate fusion. However, the implicated molecular mechanisms are strikingly different. Malleable lipids, with their ability to change shape with low energy cost, favor fusion by decreasing the overall activation energy. On the other hand, conical lipids, with their small polar head relative to the area occupied by the hydrophobic chains, tend to make fusion less energetically advantageous because they tend to migrate towards the most favorable lipid leaflet, hindering fusion pore opening. They could however facilitate fusion by generating hydrophobic defects on the membranes; this is suggested by the similar trend observed between the experimental rate of fusion nucleation and the surface occupied by hydrophobic defects obtained by molecular simulations. The synergy of dualprocess, activation energy and nucleation kinetics, could facilitate membrane fusion regulation in vivo.