We have used vesicles made from delipidated bacteriorhodopsin and synthetic lecithins to address the following questions. If the transmembrane dimension of a protein hydrophobic surface differs from the equilibrium thickness of its lipid bilayer environment, will protein monomers aggregate to decrease the protein-lipid contact surface area? If so, how large must the difference be to induce aggregation? Using lecithins with acyl chains from di-10:0 to di-24:1, the thickness of the bilayer hydrocarbon region above the lipid phase transition temperature (tm) was varied from 14.5 A less than to 7.5 A more than the transmembrane dimension of the bacteriorhodopsin hydrophobic region. Bacteriorhodopsin remains dispersed when the surrounding bilayer hydrophobic region is 4 A thicker or 10 A thinner than the bacteriorhodopsin hydrophobic surface. Only the thin- (10:0) and thick- (24:1) bilayer samples showed any bacteriorhodopsin aggregation above tm. Thus a surprisingly large difference between protein and lipid hydrophobic thicknesses can be accommodated without protein aggregation. The lipid bilayer can evidently sustain large local distortions with a small change in free energy.