A simplifying version of the lattice cluster theory is formulated for binar
y random copolymer A(x)B(1-x)/A(y)B(1-y) blends. The underlying model is ba
sed on united atom structures for the individual monomers of A and B specie
s. Applications of the theory to saturated poly(butadiene) blends display o
nly semiquantitative agreement with experiment. We discuss possible routes
to theoretical improvements and experiments that would be helpful in testin
g theory.