CFTR, whose dysfunction causes the disease cystic fibrosis, belongs to the ubiquitous superfamily of ABC proteins, which bind and hydrolyze ATP at conserved nucleotide-binding domains (NBDs). Most ABC proteins comprise membrane-spanning domains and couple ATP hydrolysis to translocation of solutes. CFTR is unique in that its transmembrane domains comprise a Cl^– channel, and ATP binding and hydrolysis at its NBDs control opening and closing of the ion-permeation pathway. Since it is likely that ABC proteins might share a common conformational coupling mechanism, linking hydrolytic cycle to conformational changes in the TMDs, we have investigated CFTR’s gating mechanism starting from the evolutionary record. In alignments of NBD sequences one can discern pairs of positions at which amino acid distribution varies in a concerted way. This correlation in evolutionary space probably reflects conservation of pairs of energetically coupled residues that mediate transmission of allosteric signals in individual ABC proteins. We selected two co-evolving positions as targets for mutagenesis, one in CFTR’s N-terminal NBD1 and the other in its C-terminal NBD2. Using several single channel kinetic parameters to characterize wild-type, each of the two single mutants, and the double mutant, we could follow how the energetic coupling between the two target positions changes during CFTR’s gating cycle. We found that the coupling energy obtained from thermodynamic mutant cycle analysis of ATP dependence of gating was not significantly different from zero, while mutation-linked changes in opening rate and open probability were less than additive, yielding coupling energies of –2.7±0.5 kT and –2.4±1.0 kT, respectively. These results are consistent with a scheme in which the two side-chains are not interacting in the closed state (when ATP binds) but become coupled as the channel opens. The finding supports the hypothesis that formation of a tight NBD1/NBD2 dimer is linked to opening of the CFTR channel pore. The use of the evolutionary record for target selection broadens the scope of the experiments and confirms the existence of a common ABC protein coupling mechanism, involving NBD dimerization.