We investigated and optimized a method for structure comparison which is ba
sed on rigid body superimposition. The method maximizes the number of struc
turally equivalent residues while keeping the root mean square deviation co
nstant. The resulting number of equivalent residues then provides an adequa
te similarity measure, which is easy to interpret. We demonstrate that the
approach is able to detect remote structural similarity. We show that the n
umber of equivalent residues is a suitable measure for ranking database sea
rches and that the results are in good agreement with expert knowledge prot
ein structure classification. Structure comparison frequently has multiple
solutions. The approach that we use provides a range of alternative alignme
nts rather a single solution. We discuss the nature of alternative solution
s on several examples.