The stearoyl-ACP delta 9 desaturase from plants is a new example of a growing number of proteins that contain oxo- or hydroxo-bridged diiron clusters. On the basis of differences in primary sequence motifs providing the cluster ligands and upon structural differences elucidated by X-ray crystallography, we now propose that the presently known, soluble diiron-oxo proteins can be grouped into two classes, I and II. Class I contains hemerythrin, myohemerythrin, and, possibly, purple acid phosphatase. Class II contains ribonucleotide reductases, bacterial hydrocarbon hydroxylases (methane monooxygenase, toluene-4-monooxygenase, and phenol hydroxylase), rubrerythrin, and stearoyl-ACP desaturases. Through the use of resonance Raman spectroscopy, we have detected symmetric (vs = 519 cm-1) and asymmetric (vas = 747 cm-1) vibrational modes in the castor stearoyl-ACP delta 9 desaturase, which are typical of oxo-bridged diiron clusters. These frequencies shift by -18 and -34 cm-1, respectively, in H218O, proving that the bridging ligand is readily exchangeable with solvent (t1/2 = 7 min). Calculation of an approximately 123 degrees Fe-O-Fe angle from the position of vs and vas and from the 18O-dependent shift in these frequencies suggests that the diiron-oxo cluster in the desaturase is triply bridged in the diferric state. In the diferrous state, the two iron sites of the cluster are structurally inequivalent, as shown by differential temperature dependence of the Mössbauer quadrupole splittings. For the class II diiron-oxo proteins, primary sequence alignments reveal conserved amino acid residues which act as iron cluster ligands, participate in a hydrogen-bonding network, and are potentially involved in O2 binding and activation. Based on this conservation, a structural model for the stearoyl-ACP delta 9 desaturase active site is proposed that has strong similarity to both ribonucleotide reductase and methane monooxygenase. However, after single turnover of the diferous state with 18O2, 18O is not detected in the oxo bridge of the castor desaturase. This is in contrast to the outcome observed for ribonucleotide reductase, suggesting the desaturase and ribonucleotide reductase differ in certain aspects of their respective O2-activation reactions.