Multidomain proteins are ubiquitous in both prokaryotic and eukaryotic proteomes. Study on protein folding, however, has concentrated more on the isolated single domains of proteins, and there have been relatively few systematic studies on the effects of domain-domain interactions on folding. We here discuss this issue by examining human ␥D-crystallin, spore coat protein S, and a tandem array of the R16 and R17 domains of spectrin as example proteins by using a structure-based model of folding. The calculated results consistently explain the experimental data on folding pathways and effects of mutational perturbations, supporting the view that the connectivity of two domains and the distribution of domain-domain interactions in the native conformation are factors to determine kinetic and equilibrium properties of cooperative folding.energy landscape theory ͉ structure-based model ͉ circular permutation O ur understanding of protein folding has been deepened by the combined efforts of experimental, theoretical, and computational studies of the last decade. Energy landscape theory describes folding as the stochastic relaxation process on the free energy surface of conformational change, where the free energy surface is determined by the compromise of conformational entropy of the polymer chain and interaction energies to stabilize the native conformation (1-3). Because proteins can fold when interactions that stabilize the native conformation dominate over the nonnative interactions that may trap the chain into the irrelevant structures, protein folding can be approximately simulated by using the interaction potentials that are derived from the knowledge of the native structure. With such structure-based models, folding of various small proteins has been simulated, and quantitative agreement between simulations and experiments has been reported (3). The agreement has been improved by simulations that further take account of the residue-dependent energetic differences (4), the atomistic packing (5-7), or the hydration structure (8), and such agreement has convinced us that the topology of the native structure is the primary determinant of the equilibrium and kinetic features of folding at least for small proteins (3) although the atomistic details perturb those features or they sometimes change the delicate balance among folding pathways (9, 10).These intensive studies on folding have been predominantly focused on small, single-domain proteins or isolated single domains of larger proteins. More than 70% of eukaryotic proteins, however, are composed of multiple domains, and hence we should ask whether the principles of folding found in single domains of proteins also apply to connected multidomain proteins as well (11). Anticorrelation between the contact order and the folding rate has been observed in multidomain proteins (12), and the structure-based simulations on multidomain proteins such as the ankyrin family (13-15) and CV-N (16) have provided consistent results with experiments. The importance of interactions...