Biotin synthase, a member of the "radical-SAM" family, produces biotin by inserting a sulfur atom between C-6 and C-9 of dethiobiotin. Each of the two saturated carbon atoms is activated through homolytic cleavage of a C-H bond by a deoxyadenosyl radical, issued from the monoelectronic reduction of S-adenosylmethionine (SAM or AdoMet). An important unexplained observation is that the enzyme produces only 1 mol of biotin per enzyme monomer. Some possible reasons for this absence of multiple turnovers are considered here, in connection with the postulated mechanisms. There is a general agreement among several groups that the active form of biotin synthase contains one (4Fe-4S)(2+,1+) center, which mediates the electron transfer to AdoMet, and one (2Fe-2S)(2+) center, which is considered the sulfur source [Ugulava, N. B., Sacanell, C. J., and Jarrett, J. T. (2001) Biochemistry 40, 8352-8358; Tse Sum Bui, B., Benda, R., Schunemann, V., Florentin, D., Trautwein, A. X., and Marquet, A. (2003) Biochemistry 42, 8791-8798; Jameson, G. N. L., Cosper, M. M., Hernandez, H. L., Johnson, M. K., and Huynh, B. H. (2004) Biochemistry 43, 2022-2031]. An alternative hypothesis considers that biotin synthase has a pyridoxal phosphate (PLP)-dependent cysteine desulfurase activity, producing a persulfide which could be the sulfur donor. The absence of turnover was explained by the inhibition due to deoxyadenosine, an end product of the reaction [Ollagnier-de Choudens, S., Mulliez, E., and Fontecave, M. (2002) FEBS Lett. 535, 465-468]. In this work, we show that our purified enzyme has no cysteine desulfurase activity and the required sulfide has to be added as Na(2)S. It cannot be replaced by cysteine, and consistently, PLP has no effect. We observed that deoxyadenosine does not inhibit the reaction either. On the other hand, if the (2Fe-2S)(2+) center is the sulfur source, its depletion after reaction could explain the absence of turnover. We found that after addition of fresh cofactors, including Fe(2+) and S(2)(-), either to the assay when one turn is completed or after purification of the reacted enzyme by different techniques, only a small amount of biotin (0.3-0.4 equiv/monomer) is further produced. This proves that an active enzyme cannot be fully reconstituted after one turn. When 9-mercaptodethiobiotin, which already contains the sulfur atom of biotin, is used as the substrate, the same turnover of one is observed, with similar reaction rates. We postulate that the same intermediate involving the (2Fe-2S) cluster is formed from both substrates, with a rate-determining step following the formation of this intermediate.