The biological membrane may be viewed as a two-dimensional solvent sys
tem, the lipid bilayer, in which the membrane components are either di
ssolved (intrinsic) or to the surface of which they are adsorbed (extr
insic). The solvent bilayer is made up of a large number of lipid chem
ical species derived from a few lipid classes. Experience with model s
ystems has shown that in mixed lipid bilayers immiscibility of compone
nts is the rule rather than the exception. This suggests that the bila
yer in a biological membrane is not a homogenous two-dimensional fluid
but rather a heterogenous system consisting of a mosaic of co-existin
g phase domains in which the phases differ both chemically and physica
lly from each other. A consequence of this is the physical separation
of membrane components, including proteins, based on their phase solub
ility. The percolation in such a phase-separated system then determine
s the range over which free lateral diffusion is possible and bimolecu
lar reactions can occur. Phase percolation and long-range translationa
l diffusion have been studied in model systems using the fluorescence
recovery after photobleaching (FRAP) technique, and theoretical work s
hows that bimolecular reaction yields can be seriously reduced in phas
e-separated membranes. Transitions between percolating and non-percola
ting states in biomembranes is proposed as a potential trigger mechani
sm in the control of membrane physiology.