The use of substances that could activate the defective chloride channels of the mutant cystic fibrosis transmembrane conductance regulator (CFTR) has been suggested as possible therapy for cystic fibrosis. Using epithelia formed by cells stably transfected with wildtype or mutant (G551D, G1349D) CFTR, we estimated the apparent dissociation constant, K(D), of a series of CFTR activators by measuring the increase in the apical membrane current. Modification of apparent K(D) of CFTR activators by mutations of the nucleotide-binding domains (NBDs) suggests that the binding site might be in these regions. The human NBD structure was predicted by homology with murine NBD1. An NBD1-NBD2 complex was constructed by overlying monomers to a bacterial ABC transporter NBD dimer in the "head-to-tail" conformation. Binding sites for CFTR activators were predicted by molecular docking. Comparison of theoretical binding free energy estimated in the model to free energy estimated from the apparent dissociation constants, K(D), resulted in a remarkably good correlation coefficient for one of the putative binding sites, located in the interface between NBD1 and NBD2.