The Mo and Fe K-edge extended X-ray absorption fine structure (EXAFS) of (Bu/sub 4/N)/sub 3/(Mo/sub 2/Fe/sub 6/S/sub 8/(SEt)/sub 9/) (1) (Et/sub 4/N)/sub 3/(Mo/sub 2/Fe/sub 6/S/sub 9/(SEt)/sub 8/) (2), and (Et/sub 3/NCH/sub 2/Ph)/sub 3/(Mo/sub 2/Fe/sub 6/S/sub 8/(SEt)/sub 3/(OPh)/sub 6/) (3) have been measured (in transmission mode) and analyzed. These clusters are known (1 and 2) or believed (3) to have the so-called double-cubane structures with two MoFe/sub 3/S/sub 4/ cubes. The interatomic distances, determined by curve fitting with theoretical phase and amplitude functions, agree well with crystallographic results. Using 1 (whose structure is most accurately known) as the model, the number of atoms neighboring the Mo and the Fe atoms can be determined to a satisfactory degree for the crystallographically disordered 2 and the structurally yet unknown 3. Curve fitting of the Mo EXAFS with a two-term expression yields average Mo-S (sulfide) and Mo-Fe distances and coordination numbers, whereas curve fitting with a three-term expression yields average Mo-S (sulfide), Mo-S' (thiolate), and Mo-Fe distances and coordination numbers. The original Mo EXAFS spectrum of 2 reported in the literature has been shown to be due to an approximately equimolar mixture of 1 and 2. The Fourier transforms of the Fe EXAFS ofmore » 1 and 2 exhibit only one peak, due to Fe-S backscattering, whereas the corresponding transform of 3 shows only two peaks, due to Fe-O and Fe-S backscatterings. The unexpected finding of the absence of backscattering due to neighboring Fe and Mo atoms is explained in terms of cancellation of the Fe-Fe and Fe-Mo waves. The apparent lack of metal-metal peak in the Fourier transforms of the Fe EXAFS of 1-3 makes these clusters inadequate as models for the Fe EXAFS of the FeMo cofactor.« less