THE nitrogenase Fe–Mo cofactor (FeMo-co)1 is usually characterised by its unique ability to restore N2-fixing ability to extracts of certain nitrogenase-less mutants1,2 and by its spectral properties, some of which can be correlated with those of the FeMo protein itself3. The role of FeMo-co in nitrogenase function is not yet clear. Recently, it was reported that uncomplemented FeMo-co catalysed BH−4 reduction of C2H2 to C2H4 in a CO-inhibited, non-ATP-dependent process4. This was taken as evidence that FeMo-co is the active site for N2 reduction and binding in nitrogenase, although conversion of N2 to NH3 by cofactor-BH−4 mixtures was not observed4. C2H2 is readily reduced by many homogeneous and heterogeneous catalysts and can give only a single 2 e− reduction product, urging caution in exclusive reliance upon it as a model substrate5,6. The observation that nitrogenase catalyses reduction of cyclopropene to both propene and cyclopropane in a ∼2:1 ratio in vitro7 and also in vivo8 suggests a more demanding criterion8 for evaluation of FeMo-co catalysis, one which still avoids the difficult energetic requirements for N2 activation and reduction. This new criterion is strengthened by the fact that cyclopropene competitively inhibits N2 (ref. 9). We find that incubation of FeMo-co with a suitable ‘apoprotein’ extract yields a reconstituted FeMo-protein activity for reduction of cyclopropene to propene and cyclopropane. FeMo-co alone, however, only catalyses cyclopropane formation in the conditions described4 for C2H2 reduction. This result and other considerations suggest that active site homology between FeMo-co and nitrogenase has not yet been fully demonstrated on a function basis.