Novel macrolides epothilones, produced by cellulolytic myxobacterium Sorangium cellulosum, have the activity to promote microtubule assembly, and are considered to be a potential successor to the famous antitumor drug taxol. The biosynthetic genes leading to the epothilones are clustered into a large operon. The multi-enzyme complex is a hetero-gene cluster of polyketide synthase (PKS) and non-ribosomal peptide synthetases (NRPS) and contains several functional modules, i.e. a loading module, one NRPS module, eight PKS modules, and a P450 epoxidase. The former ten modules biosynthesize desoxyepothilone (epothilones C and D), which is then epoxidized at C12 and C13 and converted into epothilones (epothilones A and B) by the P450 epoxidase. The NRPS module is responsible for the formation of the thiazole side chain from cysteine. The biosynthesis procedure of epothilones can be divided into 5 stages, i.e. formation of holo-ACP/PCP, chain initiation and thiazole ring formation, chain elongation, termination and epoxidation, and post-modification. The analysis of the gene cluster and the biosynthetic pathway reveals that novel epothilone analogs could not only be produced by chemical synthesis/modification, tranditional microbial technologies, but also can be genetically manipulated through combinatiorial biosynthesis approaches.